Telomeres are nucleoprotein structures found at the ends of linear chromosomes. Telomeric DNA shortens with each cell division, effectively restricting the proliferative capacity of human cells. Telomerase, a specialized reverse transcriptase, is responsible for de novo synthesis of telomeric DNA, and is the major physiological means by which mammalian cells extend telomere length. Telomerase activity in human soma is developmentally regulated according to cell type. Failure to tightly regulate telomerase has dire consequences: dysregulated telomerase activity is observed in more than 90% of human cancers, while haplo-insufficient expression of telomerase components underlies several inherited premature aging syndromes. Over the past decade, we have significantly improved our understanding of the structure–activity relationships between the two core telomerase components: telomerase reverse transcriptase and telomerase RNA. Genetic screening for telomerase deficiency syndromes has identified new partners in the biogenesis of telomerase and its catalytic functions. These data revealed a level of regulation complexity that is unexpected when compared with the other cellular polymerases. In this review, we summarize current knowledge on the structure–activity relationships of telomerase reverse transcriptase and telomerase RNA, and discuss how the biogenesis of telomerase provides additional regulation of its actions.
Telomerase is a specialized reverse transcriptase (RT) that functions primarily to maintain telomeric DNA. In humans, telomerase maintains telomeres by adding the hexanucleotide repeat TTAGGG to the 3′ ends of chromosomes. The catalytic core of telomerase is a ribonucleoprotein (RNP) composed of telomerase reverse transcriptase (TERT), the polypeptide catalytic subunit , and telomerase RNA (TER) . Unlike other cellular or viral reverse transcriptases, telomerase copies a region of its integral RNA subunit as an internal template for nucleotide addition. Thus, TERT and TER together are the minimal requirements for telomerase catalytic activity in vitro . In vivo, the composition of the telomerase holoenzyme is more complex. In addition to TERT and TER, at least four other proteins contribute to the stability and activity of telomerase RNP in human cells [4, 5]. These four proteins belong to the H/ACA family, and comprise a core heterotrimer consisting of dyskerin, Nhp2 and Nop10, and a fourth localization factor, Gar1. Another polypeptide, the WD repeat-containing protein 79 (WDR79, now commonly known as TCAB1), associates with the telomerase holoenzyme and directs its intra-nuclear localization at the Cajal body, which is important for telomere catalysis . Other factors, including 14-3-3, Hsp90, p23, pontin, reptin, NAT10, GNL3L and hnRNPs (subtypes C and U), associate with telomerase biogenesis intermediates or the telomerase holoenzyme in a manner specific to the phase of the cell cycle and telomerase catalytic cycle . These factors have transient roles in the assembly, intracellular trafficking and subcellular localization of telomerase, chaperone against degradation by nucleases and proteases, and regulate the enzyme's activity in the cell.
De novo synthesis of telomeric repeats by telomerase is temporally regulated by cell cycle-specific pathways [8-10], by protein factors in the holoenzyme complex that regulate its intracellular localization , and also mechanically through controlled telomerase access to single-stranded telomere ends [11, 12]. Telomerase is known to associate with specific members of shelterin, the telomeric DNA-binding complex, via protein–protein interactions [13, 14]. The shelterin members TPP1 and TIN2 interact with and recruit telomerase . TPP1 and POT1 align the reverse transcriptase in a catalysis-ready orientation, and facilitate addition of telomere repeats [11, 14, 16]. In human and yeast cells, the number of shelterin complexes (which is directly proportional to the number of telomeric repeats) also conveys information about telomeric DNA length to regulate telomerase access to appropriate telomere ends [13, 17]. This protein-counting mechanism provides negative feedback to direct and regulate telomerase activity at the shortest telomere ends in human cells. Telomerase catalysis is also influenced by a separate telomere binding protein complex. The human ctc1–stn1–ten1 (CST) complex binds to G-rich overhangs in competition with POT1/TPP1 . Human CST binds to newly synthesized G-rich repeats and terminates telomerase catalysis , thus permitting the switch to C-strand fill-in synthesis by lagging-strand polymerases.
TER and H/ACA proteins
Full-length mature TER is a small non-coding RNA of 451 nucleotides . The major functional structures of TER include the core domain, conserved regions 4 and 5 (CR4/CR5, also known as STE), the template boundary element (TBE), and the box H/ACA domain, which contains conserved region 7 (CR7) (Fig. 1A). Located in the 5′ portion of the TER, the core domain contains the template region and a pseudoknot. The template region, which is copied to generate the hexanucleotide telomeric repeat, consists of an 11-nucleotide sequence that may be further divided into the five-nucleotide alignment domain and the six-nucleotide template domain. The pseudoknot contains several important structural features that, when altered through mutagenesis, severely impair telomerase activity .
TER makes two independent contacts with TERT: one through the core domain , and the other through the CR4/CR5 domain . Together with TERT, the core domain of TER and the CR4/CR5 domain are the minimal requirements for reconstituted in vitro telomerase activity [21, 23]. The TBE is located in the P1b helix of TER, and restricts telomerase to copying only the six-nucleotide template domain . Alteration of either the TBE sequence or its distance from the template results in telomerase copying past the template domain.
The H/ACA domain (nucleotides 271–451) contains a hinge box (consensus sequence ANANNA, where N represents any base), and a 5′-ACA-3′ sequence at the 3′ end of mature TER. The H/ACA domain is essential for the cellular stability of TER via assembly into a ribonucleoprotein (RNP) with the four-member H/ACA–RNP complex . The members of this complex include the core heterotrimer (dyskerin, Nhp2 and Nop10) and a fourth factor, Gar1. The H/ACA proteins are necessary for proper biogenesis and assembly of TER with TERT, and they remain obligate members of the RT complex throughout its catalytic cycle. The box H/ACA structure is a defining characteristic of more than 100 known H/ACA small nucleolar RNAs (snoRNAs) and small Cajal body RNAs (scaRNAs), which help guide the sequence-specific pseudo-uridylation of pre-rRNAs and snRNAs [25, 26]. Non-coding snoRNAs facilitate the pseudo-uridylation of rRNAs, while scaRNAs guide the modification of small nuclear RNAs (snRNAs) . In addition, snoRNAs function primarily in the nucleolus, while scaRNAs perform pseudo-uridylation in Cajal bodies, dynamic sub-nuclear structures that participate in diverse cellular processes, including small RNA processing, cell proliferation, migration through the different phases of the cell cycle, and the cellular stress response .
In addition to H/ACA sequence features, all scaRNA H/ACA domains harbour the Cajal body localization signal (CAB box). Accumulation of scaRNAs in Cajal bodies depends on the presence of the CAB box, defined by the consensus sequence 5′-UGAG-3′ and found in CR7 . TER is considered a scaRNA, by virtue of its CAB box, as well as its transit through Cajal bodies at particular phases of the cell cycle, although it does not participate in pseudo-uridylation of snRNAs. Sm protein subtypes B and D3 associate with TER in a CAB box-dependent manner, and facilitate the transit of pre-telomerase TER–H/ACA-RNPs, but are unlikely to remain in the assembled telomerase holoenzyme . In contrast, the WDR79 (TCAB1) protein also binds sequence-specifically to the CAB box, and is required for successful transport of TER–H/ACA-RNP and localization of assembled telomerase to Cajal bodies . In addition to the CAB box, human TER CR7 also harbours a sequence feature for correct biogenesis and assembly of the telomerase RNP and is known as the biogenesis box (BIO box, 5′-CUGU-3′) . Located in the proximal stem loop of the H/ACA domain, the BIO box is important for mature TER accumulation  via participation in loading of two complete sets of H/ACA proteins onto the proximal and distal stem loops of the H/ACA domain. The human telomerase RNA 5′ end is protected by a trimethyl guanosine cap [32, 33], and contains guanine tracts that fold into G-quadruplex to protect against nuclease degradation. Resolution of G-quadruplex by the helicase DHX36/RHAU is important for subsequent TER maturation and telomerase assembly .
In summary, the composition of the TER–H/ACA-RNP complex identifies it as a member of the scaRNP family. As such, it is subject to the same complex regulation of biogenesis and intracellular trafficking as other endogenous scaRNPs. Despite these similarities, TER is different from other H/ACA RNAs as it is transcribed under the regulation of its own promoter instead of encoded within the introns of other protein-coding genes . TER's single-gene transcription context imparts other biological constraints, in addition to those related to scaRNP biogenesis. Specific vulnerabilities have been illustrated by studies of genetic diseases, as discussed in the accompanying reviews in this minireview series (this issue).
TERT structure–activity relationships
TERT was identified as an RT-like polypeptide through sequence analysis shortly after its discovery in Saccharomyces cerevisiae and Euplotes aediculatus [36, 37]. Since the initial discovery of these two TERTs, subsequent work has identified and characterized more than 40 TERT or TERT-like proteins in eukaryotes that, with few exceptions, share a common domain organization [38, 39]. There are three distinct domains within TERT: (a) an N-terminal extension, (b) the RT domain, and (c) a C-terminal extension (CTE) (Fig. 1B).
The N-terminus of TERT comprises approximately 400 amino acids, and may be divided into two functionally important domains: the telomerase essential N-terminal (TEN) domain and the telomerase RNA-binding domain (TRBD). The TRBD interacts with TER through its cognate RNA recognition motif [40, 41]. In humans, regions within the TRBD of TERT have an important catalytic role that is independent of its RNA-binding function . For example, the T domain of human TRBD regulates the rate of template copying during catalysis in vitro, and is only moderately related to TER binding . The TEN domain of telomerase is best known for its ability to bind to and remain associated with single-stranded telomeric DNA during catalysis [44-47]. In addition, the TEN domain contains amino acid residues that, while essential for telomerase activity and telomere maintenance in human cells, are not implicated in DNA binding [46-48]. Another region within the TEN domain known as the ‘dissociates activities of telomerase’ (DAT) region directs telomerase to the telomeric substrate in human cells [49, 50], either through proper localization of the enzyme and/or correct positioning of its catalytic site on telomeric DNA . As the TEN domain plays multiple important functions, it is not surprising that deletion of this domain abolishes telomerase activity in vitro .
The RT domain of TERT is the catalytic centre of telomerase, and contains seven universally conserved RT motifs [1, 37, 52]. Mutations of conserved residues in the RT domain cause either a complete loss or a substantial decrease of in vitro telomerase activity compared to wild-type TERT [53-57]. The RT domain may be organized in two putative sub-domains, the ‘fingers' and ‘palm’, which comprise motifs 1 to A and B’ to E, respectively . Typically, the fingers domain interacts with the nucleic acid substrate, while the palm domain contains the catalytic site [58, 59]. The fingers and palm domains of TERT are connected by a ‘primer grip’ region, which is implicated in single-strand DNA binding [60-63]. The RT domain also contains a large ‘insertion in fingers’ domain (IFD), which is typically found in TERTs but not other RTs, and is implicated in stabilizing protein–protein interactions within TERT [60, 64]. An integral feature of the RT domain of TERTs is the triad of aspartic acid residues located in motifs A and C, which are critical for telomerase's polymerase activity [53, 54, 56].
While the putative fingers and palm domains are found in the RT domain of TERT, the putative thumb domain is located within the CTE . There is weak sequence conservation in this region among TERTs, which suggests it may have a species-specific function. Mutations in the C-terminus of human TERT affect enzyme activity [65, 66], telomere length maintenance  and subcellular localization .
To date, the only complete crystal structure of the TERT molecule has been obtained from the flour beetle Tribolium castaneum . As expected, T. castaneum TERT may be sub-divided into thumb, fingers and palm domains. Comparison of T. castaneum TERT and HIV RT revealed high structural similarity with respect to the spatial arrangement of the fingers and thumb domains. The location of the active site and the structure of the nucleotide-binding pocket when bound to substrate are also highly conserved . Subsequent biochemical work to resolve the structure of T. castaneum TERT bound to a DNA/RNA hybrid as a model of telomerase RNA bound to telomeric DNA showed that TERT–RNA and TERT–DNA associations resemble those observed in HIV RT . In addition, comparison of the crystal structure of T. castaneum TERT with HIV RT  supports the role of the CTE as the ‘thumb’ domain of telomerase [44, 65], and structural analysis of T. castaneum TERT bound to a DNA/RNA hybrid identified secondary structures within the C-terminus that are implicated in stabilization of the RNA/DNA heteroduplex .
Intracellular trafficking and holoenzyme assembly
Telomerase biogenesis starts with the co-transcriptional binding of nascent telomerase RNA by the pre-formed core H/ACA trimer comprising dyskerin, NOP10 and NHP2 (Fig. 2). This pre-formed H/ACA trimer also binds to NAF1, which recognizes and binds the phosphorylated C-terminal domain of RNA polymerase II and its specific transcription factors, and mediates co-transcriptional loading of the H/ACA complex . Sequential loading of two sets of H/ACA–NAF1 complexes onto nascent TER transcripts is dependent on the H/ACA sequence motifs and the BIO box (nucleotides 415–418) . Substitution of NAF1 by the H/ACA protein Gar1 promotes nucleolar recruitment of TER–H/ACA-RNPs . Gar1-mediated TER localization at the nucleolus precedes the recruitment of Cajal body localization factor TCAB1, which promotes translocation of the TER–RNP to Cajal bodies. Association with these scaRNP factors thus plays two independent biological roles: (a) co-transcriptional binding of the core H/ACA heterotrimer protects TER from non-specific exonuclease degradation, allowing stable accumulation of TER–H/ACA-RNP, and (b) binding of H/ACA- and scaRNP-specific proteins dynamically controls the location of these complexes through binding of target-specific localization factors.
Telomerase activation is regulated at the transcriptional level through expression of TERT mRNA . The TERT polypeptide is translated by cytoplasmic ribosomes and imported back to the nucleus through the nuclear pore complex  in the presence of chaperones HSP90 and p23 . The TERT polypeptide transiently passes through the nucleolus (as inferred by its interaction with nucleolar proteins nucleolin and PinX1 ) before assembly with TER into active telomerase.
Telomerase assembly is a dynamic and highly regulated process. Populations of pre-assembled TERT complexes (without TER) and TER–H/ACA-RNPs (without TERT) may be readily detected in biochemical purifications targeting each of the holoenzyme components . Analogous to a ‘just in time’ production strategy, it is conceivable that this dynamic assembly process allows an additional level of enzyme activity regulation, and provides pools of separate TERT and TER complexes for activities that are not associated with telomeric DNA synthesis. The assembly of an active telomerase enzyme is completed by association of TERT with the TER–H/ACA RNP complex. Telomerase assembly is known to involve the ATPases pontin and reptin . TERT association with these ATPases peaks in the S phase of the cell cycle, shortly before extension of the newly replicated DNA ends. Pontin/reptin bind dyskerin and TERT independently, and thereby facilitate assembly of TERT with a TER–dyskerin (H/ACA) RNP . Following assembly of the telomerase holoenzyme, these two ATPases dissociate from the active telomerase complex. Consistent with this sequence of events, purified TERT–pontin–reptin complexes exhibit low telomerase catalytic activity, suggesting that they represent a pre-telomerase complex. Active telomerase holoenzyme is found to co-purify with known nucleolar proteins, namely NAT10 and GNL3L , implying that the active telomerase RNPs may travel between the nucleolus and Cajal bodies within the nucleus. NAT10 and GNL3L were found to predominantly associate with the catalytically active telomerase pool, although recent data suggest that TERT–NAT10–GNL3L may perform a non-canonical, chromatin-remodelling modulation activity in the absence of TER . Nonetheless, telomerase RNP co-purified with GNL3L and/or NAT10 exhibits higher specific activity when compared with telomerase RNPs not associating with these factors, suggesting that these proteins are involved in telomere catalysis.
The final level of telomerase activity regulation is its targetted delivery to the telomeres and access to the 3'OH of its primer substrate at the telomere ends. Telomeric DNA catalysis is believed take place at the Cajal bodies. Accordingly, in situ hybridization and live-cell microscopy studies showed co-localization of telomeres and telomerase at the Cajal bodies, specifically at S phase [8, 76, 77]. However, primary human cells lacking Cajal bodies are efficient in telomere length extension, arguing against the obligate requirement for Cajal body localization in telomere catalysis. In addition, studies of TER CAB box sequence mutants were found to be excluded from Cajal bodies but this had exhibited minimal effects on telomere catalysis. These data indicate that Cajal body residence, as a mechanism to concentrate low levels of telomerase RNP, may not be obligate for telomere extension. In contrast, TCAB1 contributes to telomerase catalysis through processes that involve Cajal body localization and processes that do not, highlighting the multiple functions of this polypeptide in telomerase actions and regulation [31, 78]. Finally, following recruitment of telomerase to the telomere substrate, access to single-strand telomeric DNA and the correct alignment of telomerase to telomeric DNA ends are regulated by interactions with shelterin components, as discussed in the Introduction.
Catalysis of telomere repeat synthesis
There are essentially three catalysis steps in telomerase-mediated DNA synthesis: DNA binding and positioning, synthesis of the telomeric sequence to the end of the TER template, and translocation and realignment of the catalytic site with the 3′ end of the substrate (Fig. 3). Intrinsically within the TERT sequence, the TEN domain makes multiple contacts with DNA at the 5′ end of the substrate, and is thought to ‘hold on’ to the substrate throughout repeat catalysis, providing an anchor for the enzyme. The CTE region of TERT (the ‘thumb’ domain) stabilizes the DNA/RNA hybrid, while motif E (the ‘primer grip’) positions the 3′ end of the substrate correctly in the active site . Regions within the RT domain (specifically the ‘fingers’ and ‘palm’) are implicated in interaction with incoming dNTPs.
As telomerase catalyses the addition of consecutive dNTPs to the free 3′ OH of its primer substrate, a constant number of base-paired nucleotides are maintained between the template region of TER and the primer substrate . This means that, as dNTPs are added at the 3′ end, base-pair interactions at the 5′ end are disrupted. Telomerase adds dNTPs until the 5′ TBE within TER is reached. Upon reaching the TBE, translocation of the enzyme repositions a new 3′ end in the catalytic site, and the cycle of nucleotide addition and translocation resumes.
The incorporation of individual dNTPs by telomerase is known as nucleotide addition processivity (also known as type I processivity) . After first-repeat synthesis, telomerase may either translocate on the DNA substrate and re-align TER in preparation for the synthesis of a second six-nucleotide repeat, or dissociate from the substrate completely. The propensity of telomerase to successively synthesize six-nucleotide repeats is referred to as repeat addition processivity (RAP, also known as type II processivity) and is a unique feature of telomerase amongst RTs . The ability of telomerase to synthesize multiple copies of the telomeric repeat on a single primer depends critically on several amino acid residues in TERT, as well as on structural elements within TER.
Five regions within TERT are known to contribute to RAP: the anchor site, the IFD, motif C, motif 3 and the CTE. The anchor site, found in the TEN domain of TERT, contributes to RAP by facilitating template–primer translocation. The anchor site physically interacts with the primer and prevents enzyme–substrate dissociation. In S. cerevisiae, four conserved amino acid residues within the IFD participate in template–primer translocation, possibly by stabilizing the RNA/DNA hybrid between TER and the DNA substrate . In Tetrahymena thermophilia, an amino acid residue in motif C of TERT, which contains the catalytic centre of the enzyme, facilitates RAP through its role in protein–DNA interaction . Motif 3 in the catalytic domain of human TERT participates in RAP by promoting RNA/DNA strand separation and realignment after repeat synthesis . In addition, a region in the CTE domain of human TERT contributes to RAP by an as yet unknown mechanism .
TER contributes to RAP mainly through two structural elements: the template region and the pseudoknot. In human telomerase, the length of the template region within TER controls RAP by affecting RNA/DNA base-pairing interactions during translocation [82, 83]. The pseudoknot of TER is necessary for efficient RAP in both human and Tetrahymena telomerases [84, 85]. Mutagenesis of conserved nucleotide sequences that affect formation of the tertiary structure of the pseudoknot was found to significantly perturb RAP , suggesting an important role for TER tertiary structure in telomerase catalysis. In addition to the intrinsic features of the holoenzyme, telomerase RAP is also influenced by protein factors in trans, including the shelterin components TPP1 and POT1 [14, 16].
Telomerase catalysis kinetics in human cells appears to be more complex than predicted in vitro. Telomerase acts as both a processive or distributive enzyme in vivo. As a processive enzyme, a single molecule of telomerase remains bound to a single DNA substrate, and synthesizes multiple telomeric repeats via RAP. The distributive mode of action, on the other hand, requires multiple molecules of telomerase to act sequentially on a single substrate. Whether telomerase functions as a distributive or processive enzyme in vivo is controversial. In human cancer cells, telomerase acts processively under homeostatic conditions, when length maintenance is observed . Under conditions when telomerase is over-expressed, or when cells were in recovery mode following the removal of specific telomerase inhibitor GRN163L, the enzyme was found to act distributively. The distributive mode of telomerase catalysis is associated with proliferation-related increases in telomere length. It is not clear whether the switch of catalysis mode is completely based on telomerase availability or a critical short telomere length in all human cell types, but the results provide interesting insight into telomerase function in vivo.
Recent data indicate that, in addition to telomere repair, telomerase activity and TERT expression also promote cell growth and survival through other cellular pathways. High constitutive expression of TERT in laboratory-transformed cells and cancer cell lines protects against DNA damage , regulates chromatin structure , participates in transcription co-regulation [75, 90] and affects mitochondrial functions  [see accompanying reviews in this minireview series (this issue)]. Through mutagenesis studies, these non-canonical telomerase activities were found to be functionally distinct from the enzyme's role in chromosome end-structure maintenance. This review focuses on the biogenesis and regulation of the telomere synthesis form of the telomerase holoenzyme. Biochemical studies to investigate the mechanisms and structure–activity relationships for these and other roles of telomerase that are not related to telomere synthesis are ongoing. We anticipate that data from these studies will provide clues regarding the highly complex regulation of telomerase RNP biogenesis and activation.
Over the past decade, a number of previously isolated genetic diseases were found to associate with alterations in the components of the telomerase RNP or its biogenesis pathway. The accompanying reviews in this issue on accelerated ageing diseases illustrates how the initiation and regulation of telomerase activity are crucial in normal human soma within an individual's lifespan. Haplo-insufficiency of telomerase activity may extend to include the observations that even minor perturbations of telomerase function predict disease association . Risk assessment and the possibility of telomerase replacement strategy for these diseases await complete mapping of the enzyme's telomere-related and non-telomere-related functions.
Telomerase research in the Wong laboratory is supported by the Canadian Cancer Society Research Institute (grant number 019250) and a personal award from the Michael Smith Foundation for Health Research (number CI–SCH-00102). K.R.H. is supported by a grant to CARMA (from the Canadian Institutes of Health Research, grant number FRN85515).