The chromosomes of eukaryotic organisms, which include plants, animals and fungi, are capped by protective structures called telomeres1. In certain types of proliferative cell, these structures are maintained by the telomerase enzyme. Inherited genetic mutations that compromise telomerase function cause disorders characterized by deterioration of proliferative tissues2, whereas increased expression of telomerase supports unbridled cell growth in most human cancers3. Efforts to develop telomerase-based therapeutics have been hampered by an incomplete understanding of the structure and organization of the telomerase complex. In a paper published in Nature, Nguyen et al.4 report a structure of human telomerase bound to DNA, providing an unprecedented view of how the enzyme complex is organized and establishing a framework for drug discovery.
Telomerase is a ribonucleoprotein (RNP) enzyme — part RNA, part protein. To retain normal function, the enzyme must contain at least a telomerase reverse transcriptase (TERT) protein subunit and a long, non-coding telomerase RNA (TR). Telomerase complexes also usually include several species-specific proteins that are important for regulating RNP assembly5. The enzyme maintains telomeres by catalysing the addition of simple DNA-sequence repeats onto the ends of linear chromosomes. The catalytic TERT subunit uses a small portion of TR as a template for these repeats.
Vertebrate telomerases also belong to the family of box H/ACA RNP complexes, which generally modify essential cellular RNAs such as transfer RNAs, ribosomal RNA and spliceosomal RNA6. Curiously, despite possessing the RNA and protein components of H/ACA RNPs, telomerase does not exhibit the RNA-modifying activity typically associated with this class of complex. Nevertheless, mutations in those components impair telomere maintenance and cause diseases related to telomerase deficiency2.
Human telomerase has low abundance in cells, and attempts to reconstitute it from its purified components have been thwarted by protein insolubility and the low efficiency of RNP assembly. This challenge can be overcome by expressing TERT and TR in cultured human cells that naturally express large quantities of the other telomerase proteins needed for RNP assembly7. This approach was used to reconstitute human telomerase and produce the first low-resolution structure8 of the enzyme using electron microscopy (EM). However, telomerase complexes made in this way can contain different subunits and have varying structures, both of which limit the resolution of EM structures.
Nguyen et al. report an improved purification procedure that yields substantially more-active and more homogeneously assembled telomerase. Armed with this highly purified enzyme, the authors used cryo-electron microscopy (cryoEM) to reconstruct the structure of human telomerase at subnanometre resolution. This resolution is three times better than that of the previously reported structure, and allowed the authors to reliably work out how the various telomerase components correlate with the regions of electron density mapped out by cryoEM.
The authors’ structure has a bilobal shape (Fig. 1) consistent with the earlier low-resolution structure. This shape was previously interpreted8 as a telomerase dimer — that is, each lobe was thought to contain a copy of the TERT and TR subunits. Nguyen et al. now propose a markedly different interpretation. They point out that the structure of one of the lobes is consistent with high-resolution models9 of TERT and TR, which comprise the ‘catalytic core’ of the telomerase complex. They also compare their data with the crystal structure10 of an H/ACA RNP bound to a single ‘hairpin’ RNA, and postulate that the second lobe contains the H/ACA RNP. Moreover, they propose that this H/ACA lobe includes a single copy of the TCAB1 protein, which regulates telomerase trafficking in the nucleus11. The bilobal architecture results from a remarkably extended RNA scaffold, consistent with the general model that TRs provide a flexible scaffold for telomerase-associated proteins12. The two distinct lobes interact through two bridging RNA helices.
The discrepancies between Nguyen and colleagues’ structure and previous reports of the composition and organization of human telomerase are probably rooted in the different methods used to purify the enzyme, which might yield distinct telomerase subcomplexes and, in some cases, promote formation of a telomerase dimer. Nevertheless, the present structure supports the hypothesis that telomerase from a range of organisms can function as a monomer13,14.
In Nguyen and co-workers’ reconstruction of the catalytic core of human telomerase, the active site of the ring-shaped TERT subunit contains the template RNA bound to a DNA substrate. The two regions of TR required for catalytic function in vitro are known as the template/pseudoknot domain (t/PK) and the conserved regions 4/5 (CR4/5) domain. In the authors’ structure, the evolutionarily conserved pseudoknot fold in the t/PK domain resides on the ‘back’ side of the TERT ring, far from the active site. This is highly consistent with the pseudoknot’s location in the previously reported structure13 of telomerase from the protozoan Tetrahymena thermophila, and suggests that this essential TR element indirectly promotes telomerase function, perhaps by guiding proper RNP assembly or by influencing the conformation of TERT subdomains during catalysis.
The authors’ model of the H/ACA lobe is the first structure of a eukaryotic H/ACA RNP, and includes two complete sets of the four H/ACA proteins (dyskerin, NOP10, NHP2 and GAR1) bound to RNA hairpins. The structure reveals interesting differences in the way in which each set of H/ACA proteins interacts with its respective hairpin. The foundation of the H/ACA lobe structure is made by dyskerin subunits, which bind to the base of each RNA hairpin and also make extensive contacts with each other. The remaining three H/ACA proteins seem to make numerous contacts with one of the RNA hairpins, but substantially fewer with the other. Lastly, the TCAB1 protein binds to a structure within TR known as the CAB box, found at the apex of the H/ACA lobe. These results provide a structural explanation for a wealth of previously reported biochemical data15,16 that defined the subunit stoichiometry, as well as the RNA-hairpin sequence and structure requirements for stable assembly, of the TR H/ACA RNP.
The structural segregation of the catalytic core and H/ACA lobe raises the question of why the telomerase complex has retained the large H/ACA RNP throughout evolution. Part of the answer is likely to be that TCAB1 is needed to mediate trafficking of telomerase components during RNP assembly. However, another striking feature of the H/ACA lobe is that its proteins seem to promote a specific conformation of TR that brings the CR4/5 domain close to the catalytic core, where it is needed to support catalysis. This bears a remarkable resemblance to the T. thermophila telomerase RNP, in which an RNA-binding protein called p65 is required to remodel TR and promote contacts with the catalytic TERT subunit13,17. This resemblance suggests that telomerases from different organisms have co-opted different types of RNA-binding protein to promote RNP assembly. The idea that the H/ACA proteins direct RNA folding can be tested experimentally, using the new structural model as a framework with which to design and interpret incisive experiments.
Nguyen and colleagues’ work sets a new bar for structural models of human telomerase, but there is still work to be done. A complete view of telomerase structure and function will require atomic-resolution models of the enzyme in its various functional states. Advances in cryoEM techniques, together with other structural and biophysical tools, will no doubt continue to enhance our appreciation of the intricate workings of this fascinating enzyme.
Nature 557, 174-175 (2018)