|Home | About | Journals | Submit | Contact Us | Français|
Chromosome stability requires a dynamic balance of DNA loss and gain in each terminal tract of telomeric repeats. Repeat addition by a specialized reverse transcriptase, telomerase, has an important role in maintaining this equilibrium. Insights that have been gained into the cellular pathways for biogenesis and regulation of telomerase ribonucleoproteins raise new questions, particularly concerning the dynamic nature of this unique polymerase.
Linear chromosome ends, or telomeres, must be distinguished from sporadic DNA breaks so that only broken ends are rejoined by repair. Telomeres of eukaryotic nuclear chromosomes typically harbour an array of simple-sequence repeats. These telomeric repeats were first characterized in the macronucleus of the ciliate Tetrahymena thermophila as ~50 copies of 5′–T2G4–3′ (REF. 1). In vertebrates, telomeres have ~1000 copies of 5′–T2AG3–3′ (REFS 2,3). Telomeric repeat DNA assembles with numerous proteins to create an end-protective chromatin structure (reviewed in REFS 4–7).
Sequence erosion of terminal repeats is inherent in each round of genome replication. The replenishment of telomeric repeats can occur by recombination-based mechanisms, but it more commonly involves the extension of the 3′ ends. This de novo extension reaction is mediated by the essential eukaryotic reverse transcriptase (RT), telomerase. The telomerase ribo-nucleoprotein (RNP) carries an internal template for telomeric repeat synthesis and specifically recognizes chromosome ends as substrates. Telomerase-mediated repeat synthesis can balance the loss of chromosome ends that occurs during genome replication and damage to maintain the homeostasis of telomere length. By contrast, if this balance is not maintained, cumulative loss or gain of telomere length eventually perturbs the insulating function of telomeric chromatin. Typically, compromised telomere structure halts cell proliferation as a response to DNA damage4. Othewise, cell division without sufficient telomeric repeats induces large-scale genomic instability8,9.
The important role of telomerase activity in the maintenance of genome integrity suggests that telomerase should be a constitutive ‘housekeeping’ enzyme. Instead, it has been shown that telomerase is extensively regulated. Multicellular organisms have developmental, tissue-specific and stress-responsive strategies for telomerase repression10–12. The inactivation of telomerase and telomere-length maintenance in human somatic cells has been proposed to function as a tumour-suppressor mechanism13–15. Inactivation of telomerase might also be required for quiescence, differentiation and death of some cell types16. However, cumulative telomere erosion limits the renewal capacity of highly proliferative human cell lineages in the skin and blood17,18. Active telomerase has been detected in extracts of human cells from early embryogenesis, the germline, a subset of epithelial and lymphoid progenitors and almost all types of cancer12,17. Because cancer cells strongly upregulate telomerase activity and remain dependent on telomerase function for viability, telomerase inhibitors are under development as potential anti-cancer therapeutics8,19.
Telomerase studies address questions with broad relevance to diverse fields of current research. Recent molecular and biochemical studies of telomerase have focused on the properties of the enzyme as a specialized polymerase, a dynamic RNP complex and a factor that binds to specific regions of each chromosome. This review aims to draw together insights on telomerase in its physiological context, with emphasis on molecular and biochemical characterization across three model systems: budding yeast, vertebrates and ciliates. After introducing telomerase, successive sections describe the requirements for endogenous RNP biogenesis, the requirements for activity on telomeres, and last, other regulatory factors that are involved in telomerase activity. The review concludes with a discussion on enzyme structure and mechanisms of activity. It is important to refine and broaden the approaches that are used to study telomerase in the physiological context due to direct implications of telomerase deregulation and deficiency in human disease18,20,21.
RTs are best known as viral proteins that copy an RNA genome into DNA. The discovery of a eukaryotic RT occurred in the pursuit of an understanding of telomere dynamics. The first studies of telomere structure were performed in ciliated protozoa, which share a developmentally regulated process of telomere synthesis at formerly non-telomeric sites of chromosome fragmentation22. A similar process was experimentally induced in Saccharomyces cerevisiae by introducing plasmids with ciliate-chromosome termini that gained extra caps of the distinct yeast telomeric repeat23. Cell extracts from the ciliate T. thermophila yielded biochemical evidence of an enzyme that could extend a single-stranded primer with single-stranded telomeric repeats24. Initially, the persistence of this activity in micrococcal nuclease-treated extracts led to this enzyme’s designation as telomere terminal transferase24, with potential analogy to the sequence-specific nucleotide-transferase CCA-adding enzyme, which completes or restores the 3′ end of mature transfer RNAs25. Subsequently, the repeat-synthesis activity was recognized to be nuclease-sensitive in a unique way. The enzyme, now termed telomerase, required a specific RNA molecule with an internal sequence that could be used as a template for the synthesis of telomeric repeats26,27. RNase-sensitive synthesis of telomeric repeats was next shown in cell extracts from the divergent spirotrichous ciliates Oxytricha nova and Euplotes crassus28,29, and then leapt onto the main stage following its detection in extracts from cultured human cancer cells30.
Pioneering molecular genetics studies in T. thermophila established that the integral telomerase RNA subunit (TER) provides the template for repeat synthesis at chromosome ends31,32. Curiously, there are exceptions to this rule. Repeat synthesis in some organisms, such as the ciliate Paramecium tetraurelia, is not strictly determined by the template33. Also, repeat synthesis can variably fail to extend for the entire length of the template, as occurs in S. cerevisiae34,35. Our current knowledge of the primary sequence of TER is restricted to yeasts, ciliates, vertebrates and one virus36,37.
The catalytic activity of telomerase requires TER and the protein telomerase RT (TERT). TERT harbours a central region that has homology with the active-site motifs of the viral RT enzymes38–40. Although recombinant TERT cannot extend a typical RT substrate of annealed template and primer41, it can catalyse a weak nucleotide-transfer reaction42. Many organisms encode a single isoform of TERT. However, E. crassus has three genes that encode distinct TERT proteins and show differential regulation that is correlated with changes in telomerase activity 43. Also, human TERT pre-mRNA is alternatively spliced in a manner that is predicted to alter telomerase function at telomeres44. The biological significance of these examples of isoform diversity remains to be tested experimentally.
In telomerase-sized RNP enzymes, such as eukaryotic RNase P, multiple proteins assemble with each other and their RNA in a tightly coupled manner45. The depletion of any subunit of the endogenously assembled RNase P holoenzyme affects RNP accumulation and/or its activity in cell extracts. By contrast, evidence from studies in yeasts, vertebrates and ciliates strongly indicates that endogenous assembly of telomerase holoenzymes is more complicated. The physiological pathways of telomerase biogenesis and regulation generate a diversity of complexes that contain TER and/or TERT. Some activities of TER and TERT have been proposed that are independent of telomere synthesis, and in some cases independent of each other16,46–49. In this article, the term ‘telomerase RNP’ will be used to describe all endogenously assembled complexes that contain TER. The term ‘telomerase holoenzyme’ will be used to describe the subset of these RNPs that also contain TERT. However, this working definition of telomerase holoenzyme might need to be reconsidered as we gain new insights into the dynamics of telomerase complexes in vivo.
Purified TER and TERT do not form active RNP. However, TER and TERT can be combined in the presence of a crude eukaryotic cell extract such as rabbit reticulocyte lysate to generate catalytically active RNP50. Similarly to many other RNP enzymes, telomerase relies on a step-wise cellular RNP biogenesis pathway to provide assembly specificity. As cells differ in their levels of telomerase RNP depending on growth conditions and cell type, this non-spontaneous process of endogenous telomerase RNP biogenesis also provides an opportunity for enzyme regulation.
Tightly associated TER-binding proteins that are required for the biological stability of telomerase RNP have been identified. These proteins are mandatory subunits of an endogenously assembled telomerase RNP, and they constitute the structural core of any telomerase holoenzyme. Dissociation of these proteins from TER is expected to induce RNP turnover. Notably, TERT is not required for the biological stability of yeast and vertebrate telomerase RNPs. In yeast, telomerase RNPs accumulate in vivo in the presence or absence of co-expressed TERT51. Similarly, human telomerase RNPs accumulate ubiquitously in cells regardless of the presence of telomerase activity in cell extracts52.
Several of the cellular requirements for generating a biologically stable telomerase RNP in budding yeast are similar to the requirements for the assembly of spliceosomal small nuclear (sn) RNPs53,54. The yeast TER primary transcript is synthesized by RNA polymerase II, polyadenylated, capped on its 5′ end and processed to remove the polyadenosine tail34,53 (FIG. 1a). Sequence inspection suggested the presence of a uridine-rich consensus motif that is required for association with Sm proteins54. Sm proteins were originally identified as human autoantigens and were later shown to co-assemble into a hetero-heptameric complex that provides the biological stability of many snRNAs55. Mutation of the consensus Sm-binding site in S. cerevisiae TER or the genetic depletion of the Sm D1 protein drastically reduces telomerase RNP accumulation54. Importantly, epitope-tagged Sm D1 co-purifies the active telomerase in cell extracts, which is confirmation of an association with telomerase holoenzyme. Additional aspects of the yeast telomerase RNP biogenesis pathway remain to be investigated, including whether the shuttling of telomerase RNP between the nucleus and the cytoplasm56,57 is important for the specificity of the assembly pathway.
Studies of human TER led to the first insights about the endogenous pathways of telomerase RNP biogenesis and brought attention to the role of telomerase deficiency in human disease (BOX 1). The human TER primary transcript is synthesized by RNA polymerase II, capped on its 5′ end, internally modified, and processed at its 3′ end to generate the mature, functional TER17,52,58 (FIG. 1b). The proper processing and stability of human TER are dependent on a motif that is formed by the 3′ half of the molecule59. This H/ACA motif is common to vertebrate TER and conserved RNA families that guide the post-transcriptional modification of non-protein-coding RNAs60. In vivo studies of general H/ACA-motif RNP assembly have shown the co-transcriptional binding of three proteins essential for RNP accumulation — dyskerin (Cbf5 in yeast), NHP2 and NOP10 — and the subsequent joining of a fourth protein, GAR1 (REF. 60). Human TER also assembles with these four H/ACA-motif RNP proteins. Importantly, endogenous and epitope-tagged versions of these proteins co-purify active telomerase from cell extract, as expected for an association with telomerase holoenzyme61–64.
Telomerase deficiency in humans was first described in the disease dyskeratosis congenita (DC). Patients with DC share signs of insufficient cellular renewal in the skin and blood, including nail loss and low blood-cell counts, and they die primarily from bone-marrow failure20,21. The predominant X-linked inheritance of DC arises from substitutions in the RNA-binding protein dyskerin. Dyskerin, NHP2, NOP10 and GAR1 (which are all involved in RNP biogenesis) assemble together on H/ACA-motif RNAs to form ribonucleoproteins (RNPs) that modify ribosomal RNAs and other functional RNAs by conversion of uridine to pseudouridine60. The H/ACA-motif RNAs determine the sites of modification by hybridization to the target sequences. Dyskerin, NHP2 and NOP10 are essential for the cellular accumulation of the H/ACA-motif RNAs. Also, dyskerin contains the active site for the modification reaction. Cells from patients with X-linked DC that express variants of dyskerin have normal levels of H/ACA-motif RNAs and pseudouridine modifications, but accumulation of the telomerase RNA subunit (TER) is reduced61,152. Telomerase deficiency leads to premature telomere shortening, which in turn limits the renewal capacity of highly proliferative cell types in skin and blood.
The less common, and generally less severe, autosomal dominant inheritance of DC was next investigated, and the disease was shown to coincide with the heterozygous expression of a TER variant153. Shortly thereafter, patients with aplastic anaemia were also shown to have heterozygous expression of a TER variant154. Mutations in the gene that encodes telomerase reverse transcriptase (TERT) have recently been detected in patients with blood diseases155. These, and potentially other, distinct human disorders of cell proliferation can all be derived from telomerase deficiency, yet they differ in clinical presentation because the extent and the mechanism of telomerase inhibition affects the spectrum of tissue involvement17,18,20,21.
Although the consensus H/ACA motif is necessary and sufficient to direct RNP assembly for most H/ACA-motif RNAs, it is not sufficient for the assembly of a telomerase RNP that retains full-length TER59,65. The extra requirement for TER accumulation involves the loop of the H/ACA-motif 3′ hairpin, which is proposed to interact with an unknown RNP biogenesis factor64,65. This 3′ loop has two separable roles; it is involved in RNP biogenesis and in RNP enrichment in subnuclear compartments that are known as Cajal bodies58,66. The observed enrichment of telomerase RNP in Cajal bodies is not required for RNP biogenesis or catalytic activation58,66, and so the biological significance of this localization remains to be determined. It will be important to gain insight into the missing details of the vertebrate telomerase RNP biogenesis pathway for two reasons. If additional TER-specific protein–RNA interactions are discovered, they can be used to develop screens for clinically useful telomerase inhibitors. Also, knowledge of the vertebrate-specific features of telomerase RNP biogenesis might provide clues leading to the discovery of vertebrate-specific strategies for the regulation of enzyme functions in vivo.
Ciliates have a unique pathway for telomerase RNP biogenesis. The ciliate TER primary transcript is synthesized by RNA polymerase III and not processed, leaving a 3′ tail of polyuridine that was added during transcription termination27 (FIG. 1c). Oligonucleotide-based telomerase affinity purification from Euplotes aediculatus recovered TERT and a 43 kDa protein (p43) with a divergent La motif for RNA-binding67,68. On the other side of the ciliate family tree, T. thermophila telomerase affinity purification using endogenously expressed epitope-tagged TERT identified several holoenzyme proteins, including one of 65 kDa (p65) with the same divergent La motif 69. Importantly, antibodies against p43 or p65 deplete telomerase activity from cell extract, and epitope-tagged p65 co-purifies telomerase activity, which demonstrates the association of these proteins with telomerase holoenzymes68,69. Recombinant p43 or p65 binds directly to TER, with a requirement for the base-paired stems of TER rather than the single-stranded regions that are a typical requirement for La-motif-protein binding70–72. Genetic depletion of T. thermophila p65 drastically reduces TER and TERT accumulation without affecting other RNPs69, which indicates that this protein has an essential role in the cellular biogenesis of stable telomerase RNP.
Results from recent studies with T. thermophila p65, TER and TERT offer a rationale for the evolution of multi-step telomerase RNP biogenesis pathways. In an entirely recombinant system, the p65–TER interaction greatly stimulates the TERT–TER interaction through a conformational change in the bridging TER71,72. This finding indicates that a tightly associated TER-binding protein functions both to stabilize telomerase RNP and to fold TER into a structure that is favourable for TERT binding and enzyme function. In the future, it will be important to use direct structural assays to determine p65-mediated changes in TER conformation and to test the generality of these findings for telomerase RNP biogenesis in other organisms.
Mechanisms that govern telomerase interaction with telomeres have been studied mostly from the telomere side73. To accomplish end protection, a telomere should deny enzymes access to the 3′-OH end of the chromosome (FIG. 2, end-protected telomere). Replication and processing reactions are proposed to open the telomere to potential remodelling (FIG. 2, open telomere). The open telomere can then attract enzymes or binding proteins that generate a structure that allows elongation by telomerase (FIG. 2, telomerase-enabled telomere). Within a cell, each telomere must function in an autonomous fashion to solicit more or less lengthening, depending on the differential of its own length from the homeostasis.
It is less obvious why the engagement of telomerase with the telomere is regulated from the telomerase side. This level of regulation has been particularly well established in the physiological context of blood cells, in which some ‘telomerase-positive’ cell populations still suffer telomere erosion17,74. Perhaps telomerase-subunits need to be released from sites of safe storage (FIG. 2, inhibitory sequestration). Or, a change in RNP conformation or composition could be required to gain full DNA-binding specificity (FIG. 2, telomere-enabled holoenzyme). Alternatively, factors that are required to dissociate telomerase from the telomere could be over-active (FIG. 2, right half). In addition to these protein–protein and protein–DNA interactions, telomerase interaction with the telomere might also be regulated by DNA–DNA interactions (BOX 2).
The first genes that were shown to have roles in telomere-length maintenance were identified in S. cerevisiae. Mutant strains with defects in telomere replication were isolated by screening for several phenotypes, including progressive telomere erosion, which led to the identification of the ever shortening telomere (EST) genes75,76. EST2 encodes TERT38 and EST4 encodes a partial loss-of-function allele of the telomere-binding protein Cdc13 (REFS 76,77). The EST1 and EST3 genes encode proteins that function only, at least by genetic criteria, in telomerase-dependent telomere-length maintenance. Strains that lack Est1 or Est3 show proliferation-dependent telomere erosion, yet they retain telomerase activity in cell extracts51. Importantly, epitope-tagged Est1 and Est3 co-purify active telomerase from cell extract, as is expected for holoenzyme proteins78. Est1 associates with TER independently of TERT, probably as a direct interaction (FIG. 1a), whereas Est3 associates with TER only through TERT78–80.
Results from directed interaction assays and chromatin immunoprecipitation experiments indicated that there are at least two mechanisms for the physical recruitment of telomerase to the telomere77,81. The first interaction, which is correlated with telomere elongation, occurs in the S phase of the cell cycle and involves Cdc13 (REF. 82). A second interaction, which is evident in the G1 phase, is mediated through the association of TER with Ku, a DNA-end-binding protein81. Ku is not essential for telomere maintenance, but the disruption of the direct Ku–TER interaction reduces telomere length and reduces chromosome healing at DNA breaks83. The factors that discriminate these two modes of telomere association are not obvious, as Ku binds to TER at a site that does not overlap with other known protein–TER interactions80,84 (FIG. 1a). Est1 levels rise on S-phase entry and drop with S-phase exit82, so it might be that either an excess of Est1 triggers the holoenzyme’s S-phase mode of engagement or that the removal of Est1 from the holoenzyme forces it out of the S-phase mode.
The mammalian heterogeneous nuclear (hn) RNP proteins A1 and C have been proposed to link the telomerase holoenzyme to telomeres, potentially by bridging TER and single-stranded DNA85,86. These and other abundant hnRNP proteins package a broad spectrum of RNA polymerase II transcripts, so it is therefore difficult to establish a function for hnRNPs that is specific to the context of the telomerase holoenzyme. Recent studies indicated that hnRNP A1 stimulates telomerase activity by counteracting single-stranded product DNA folding87. Other factors that might link the telomerase holoenzyme to telomeres include three vertebrate proteins that have been characterized based on limited homology with the budding and fission yeast Est1 (REF. 88). Two of these factors have been described as holoenzyme proteins89,90, but as these proteins have important functions beyond the regulation of telomerase activity, their roles specific to the context of the holoenzyme might be difficult to establish.
In T. thermophila, subunits of the telomerase holoenzyme affect the ability of the enzyme to extend chromosome substrates (FIG. 1c). The genetic depletion of the p45 telomerase protein induces telomere shortening without affecting RNP accumulation in vivo69, which is similar to the null phenotype of EST1 and EST3 knockouts in budding yeast. Importantly, antibodies against T. thermophila p45 deplete telomerase activity from cell extract, and epitope-tagged p45 co-purifies active telomerase, which confirms the association of p45 with the telomerase holoenzyme69. It will be interesting to determine and compare the mechanisms by which the S. cerevisiae and the T. thermophila holoenzyme proteins activate telomerase for function at telomeres, because these two single-celled eukaryotes have quite distinct telomere and telomerase structures.
Single-stranded telomeric repeat DNA that contains tracts of G nucleotides can fold to form stable intramolecular or intermolecular G-quartet structures. G nucleotides in a planar configuration form Hoogsteen base-pairs, each constituted from one canonical and one non-canonical pairing edge of the base. Four telomeric repeats donate G-tracts such that consecutive Gs within a repeat are stacked and pair in successive quartets. G quartets generally reduce the efficiency of primer extension by telomerase in vitro156,157.
Does this type of DNA folding regulate the interaction of telomerase with endogenous telomere substrates? In spirotrichous ciliates, telomere-binding proteins have been proposed to promote G-quartet formation158. In other organisms, G-rich-strand folding is thought to be actively resolved to prevent accelerated telomere shortening159. It remains to be established whether G quartets or other structures that are formed by G-rich single-stranded telomeric repeat DNA have a direct role in modulating telomerase activity at telomeres.
The holoenzyme proteins that are required for telomerase RNP accumulation, activity and function at the telomere make essential contributions to telomere maintenance. Although the characterization of telomerase-associated proteins is far from complete, numerous additional factors have been suggested to interact with the telomerase holoenzyme and/or intermediates in its biogenesis and regulation. Among their many roles, these factors probably determine the dynamic distributions of TER and TERT within a cell. Particularly in vertebrate cells, TER and TERT localizations have been described to vary across the cell cycle without mutual coordination66,91,92. This dynamic regulation could be partially directed by cellular chaperones that are required for holoenzyme assembly, disassembly and degradation. The physiological significance of the dynamics of telomerase subunits will be important to investigate in future studies.
Human TER accumulates in Cajal bodies66,91–93, which are readily detectable in a subset of cell types94. TER also associates with nucleoli59,92 and colocalizes with a subset of telomeres in S phase91,92. Recombinant human TERT can be detected in the nucleoplasm, nucleoli and Cajal bodies93,95–99, whereas endogenous TERT has been localized to the nucleoplasm, nucleoli, nucleoplasmic foci and telomeres92,99–101. TERT localization can change in response to DNA damage or the expression of an oncogene98. TERT also seems to be redistributed from the nucleus to the cytosol or vice-versa; this translocation correlates with telomerase activation, cellular stress and/or protein phosphorylation102–105.
TER and the p43 holoenzyme protein from O. nova or E. aediculatus accumulate in dispersed macronuclear foci that are separated from chromatin, and only a small percentage of this pool is transiently recruited to zones of DNA replication70,106,107. Euplotes crassus TERT localizes transiently to replication bands without obvious concentration in foci108, whereas E. aediculatus TERT shows more dispersed macronuclear localization107. In yeast, the endogenous levels of TER and TERT have not been detected using cytological methods. Overexpression studies have shown that TER accumulates in the nucleoplasm, whereas TERT accumulates in the nucleolus56,57. However, their co-overexpression shifts the localization of TERT into the nucleoplasm with TER. In yeast, the nucleolar localization of TERT that is not associated with TER is consistent with competition for TERT binding between TER and the nucleolar ribosome-biogenesis protein PinX1 (REF. 109). In comparison, human PINX1 has been proposed to bind to and inhibit the activity of the telomerase holoenzyme110,111.
Transient associations of TER and TERT with chaperone activities occur during the assembly, disassembly and degradation of telomerase complexes. Based on well-established precedents for the assembly of Sm proteins and H/ACA-motif RNA-binding proteins55,60, the pathways for telomerase RNP biogenesis in budding yeast and vertebrate cells must involve the participation of RNP-assembly chaperones. The heat-shock protein-90 (HSP90) and p23 chaperones associate with human TERT, and the HSP90 inhibitor geldanamycin reduces recombinant or endogenous telomerase activity that is assayed in vitro112,113. Geldanamycin also induces TERT ubiquitylation and degradation by the proteasome, which is mediated by the E3 ubiquitin ligase MRKN1 (REF. 114).
With chaperones, it can be difficult to discriminate whether an association reflects an involvement in the normal assembly or disassembly pathway of a complex, or an involvement in the recycling or disposal of a misfolded, off-pathway complex. The T. thermophila p80 and p95 proteins and vertebrate TEP1, which were originally characterized in association with TER and telomerase activity, are much more abundant than telomerase and are genetically dispensable for telomerase function in vitro and in vivo69,115–117. Based on the bioinformatic prediction of p80 homology with Ro118, a protein that is thought to detect misfolded RNA or RNP in vivo119, p80 could have a general function in RNP recycling or degradation. Telomerase could be a physiological target of regulation by RNP recycling or degradation, which potentially accounts for telomere lengthening in T. thermophila strains that lack p80 and/or p95 (REF. 115). However, that p80 and/or p95 interact with TER under disruptive purification conditions is more consistent with RNP rearrangement in cell extract69. Likewise, cell-extract-induced protein–RNA interactions could account for the associations of human TER and telomerase holoenzyme with abundant RNA-binding proteins such as La, Staufen, and the ribosomal protein L22 (REFS 120,121). In vivo crosslinking assays to test RNP composition prior to cell lysis122 have been developed for studies of more abundant RNPs. It will be important to apply these assays to studies of telomerase as they could provide insights into the physiological interaction partners of telomerase holoenzymes.
Compared to nucleic-acid synthesis that is performed by other polymerases, telomerase-mediated telomeric repeat synthesis requires a more elaborate coordination of template, substrate and active site. The internal template must be brought to the active site in a manner that excludes the regions beyond the 5′ and 3′ template boundaries. The substrate must engage sequence-specific single-stranded-DNA-binding sites to give only its extreme 3′ end the potential to pair with the template. With each successive nucleotide addition, architectural changes of the template and product in relation to the active site must occur (FIG. 3; see legend for details). To keep these moving parts in coordination, strong interactions that remain fixed relative to the active site should anchor the template and substrate in the RNP complex (FIG. 3; BOX 1).
Telomerase holoenzymes from different organisms, or from different developmental stages of the same organism, generate surprisingly diverse profiles of product synthesis123,124. Little is known about which features of telomerase activity in vitro affect its function at telomeres in vivo. Is it significant that relatively long products are synthesized by human and T. thermophila holoenzymes, whereas short products are synthesized by the mouse and yeast holoenzymes? Is the structure of the chromosome end influenced by the phylogenetic variation in the repeat permutation that is generated by synthesis to the template 5′ end? To address these questions, more knowledge is required about the structures and mechanisms that direct repeat synthesis. To date, high-resolution structures have been determined only for isolated motifs of TER and for an N-terminal domain of T. thermophila TERT125,126.
The internal template is defined not by its own sequence but instead by flanking regions that establish the 5′- and 3′-template boundaries. The T. thermophila TER 5′-template boundary is determined, at least in part, by tight binding of a template-flanking motif to TERT127. In other organisms, the region of TER that is adjacent to the 5′ end of the template forms a base-paired duplex that is required for maintaining the 5′ template boundary128,129. These TER–TERT and TER–TER interactions could occur in combination to reinforce the necessary halt of DNA synthesis at the 5′ end of the template. How the 3′ end of the template is defined is less clear. Elongation of an entirely non-telomeric-sequence primer by ciliate telomerases is mediated through the addition of a precise permutation of the telomeric repeat, which indicates that the template has a default position in the active site. The 3′-flanking region of the T. thermophila TER template can bring an exchangeable oligonucleotide template to the active site130. It is therefore likely that at least transient TER–TERT or TER–TER interactions 3′ of the template help to position the template 3′ end. To correctly extend 3′ substrate ends paired in variable register with the template, the template must be able to move variable distances from the default position by forming a template–primer duplex.
The mechanisms that underlie the recognition of a single-stranded-DNA substrate seem surprisingly variable between holoenzymes and have not yet been well characterized123,124. Activity assays, interaction assays and high-resolution structure have defined a binding surface for single-stranded DNA that is adjacent to the template hybrid, termed PAS1 (primer/product alignment/anchor site-1). DNA interactions with PAS1 (FIG. 3) are proposed to contribute to primer-binding affinity, alignment of the substrate 3′ end and a limited extent of repeat-addition processivity125,131–133. The extensive characterization of the primer preferences of ciliate and vertebrate telomerase holoenzymes indicates that these telomerase enzymes have additional DNA-interaction specificity (PAS2; FIG. 4b). PAS2 sites could be contiguous with or separated from PAS1 and are proposed to account for the enhanced binding affinity of longer primers, high repeat-addition processivity and the efficient elongation of primers with a 5′-telomeric sequence seed and non-telomeric 3′ end123,124.
TERT domains (FIG. 4a) have a mostly unknown architecture, so only a speculative working model for their coordination can be proposed (FIG. 4b). The TERT high-affinity RNA-binding domain (TRBD) is sufficient for interaction with T. thermophila TER in vitro or human TER in vivo134. Bacterially expressed T. thermophila TRBD does not require chaperones for assembly with TER, which indicates that the chaperone requirement of full-length TERT might reflect a need to expose, rather than fold, the TRBD135. Additional sites of TERT–TER contact have been proposed for T. thermophila and human TERTs. However, these sites have relatively low binding affinity and it is therefore difficult to study the specificities of their interactions.
The TERT essential N-terminal (TEN) domain is highly conserved among vertebrate proteins, which is an exception to the generally low level of sequence conservation that occurs beyond the active site. Activity assays and high-resolution structure analysis implicate the T. thermophila TEN domain as a PAS1 contact (FIG. 4b). The TEN domains of yeast and human TERTs could provide PAS1 contacts as well125,136–138. It is tempting to speculate that the T. thermophila TEN domain associates with the 3′-flanking region of the template (FIG. 4b) for at least part of the catalytic cycle135,139. The TERT C-terminal extension (TEC) is not essential for yeast holoenzyme function in vivo140. However, substitutions of the human TERT C terminus abrogate its physiological function in the same manner as some substitutions in its TEN domain141,142. One plausible model to account for these observations is that an interaction between TEC and TEN could provide PAS2 (FIG. 4b).
Tetrahymena thermophila telomerase RNPs assembled in vivo or in vitro are purified in complexes that contain a single TERT and a single TER69,143,144. The S. cerevisiae telomerase holoenzyme also purifies from extract without evidence for subunit multimerization, unless single-stranded DNA has been added to the extract145,146. Human telomerase RNPs reconstituted in vitro are purified as interdependent multimers of TER and TERT, with functional interference from the co-assembly of mutant and wild-type TER templates147. Human telomerase RNPs that were reconstituted in vivo under conditions of wild-type and mutant TER co-expression did not show this type of functional interference65,148–150. Notably, the TEN domain of one TERT polypeptide and the TRBD, RT, and TEC domains of another can complement each other in function to generate enzyme activity in vitro151. Perhaps the human TERT TEN and TEC domains form an intramolecular interaction during their assembly with TER in vivo, whereas in vitro they assemble preferentially by intermolecular complementation. Human TERT multimerization has been proposed to account for the large mass of the endogenous holoenzyme, which even after some disassembly exceeds the expected mass for a single TERT and TER alone147. This additional mass can instead be constituted by the holoenzyme proteins that direct telomerase accumulation and function in an endogenous context.
It is now becoming clear that there are intricate pathways of telomerase assembly and regulation. First, the biogenesis of the telomerase RNP proceeds through sequential steps of protein–RNA interaction. Budding yeast, vertebrates and ciliates use different TER-binding proteins to produce a biologically stable telomerase RNP. The assembly of RNP with TERT then leads to RNP activation, which seems to be a dynamic process. Additional holoenzyme proteins give the enzyme an ability to function on chromosome substrates. Last, TERT, TER and single-stranded DNA form a network of protein–nucleic-acid interactions that orchestrate the proper positioning of the template and primer in the active site.
Knowledge that has been gained over the past decade has highlighted the questions that remain to be answered. The unexplained difference in results from reconstitutions of TERT and TER in vitro and in vivo indicates a gap in our understanding of their interaction specificity. How does endogenous telomerase biogenesis bring the inactive RNP and TERT together? Is this assembly step reversible and is disassembly used to regulate the level of enzyme activation? Another important direction of future progress will be to understand the mechanisms that provide specificity for telomerase elongation of primer substrates in vitro and chromosome ends in vivo. How do holoenzyme proteins stimulate activity on telomere substrates? Do they induce a conformational change that locks the enzyme onto DNA or do they bind to DNA directly? Insights into the cellular requirements for telomerase biogenesis and regulation will hopefully lead to improved telomerase reconstitution in vitro. Also, studies of recombinant telomerase will provide more insights into the workings of an enzyme that is uniquely devoted to repeat synthesis.
I apologize to authors whose work could not be described thoroughly within the text and reference limits. Research in the Collins laboratory has been supported by the National Institutes of Health, Burroughs Wellcome Fund, American Cancer Society and the UC Cancer Research Coordinating Committee. I thank the members of the Collins laboratory over the past decade for creating such a stimulating and supportive research environment. I also thank T. Cech and C. O’Connor for their comments on the original draft of this review.
Competing interests statement
The author declares no competing financial interests.
The following terms in this article are linked online to:
Entrez Gene: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene.
EST1 | EST2 | EST3 | EST4
Cbf5 | dyskerin | GAR1 | TERT | NHP2 | NOP10
Kathleen Collins’ homepage: http://mcb.berkeley.edu/faculty/BMB/collinsk.html
Access to this links box is available online.