The enzyme telomerase adds telomeric repeats to chromosome ends to balance the loss of telomeres during genome replication. Telomerase regulation has been implicated in cancer, other human diseases, and ageing, but progress towards clinical manipulation of telomerase has been hampered by the lack of structural data. Here we present the cryo-electron microscopy structure of the substrate-bound human telomerase holoenzyme at subnanometre resolution, showing two flexibly RNA-tethered lobes: the catalytic core with telomerase reverse transcriptase (TERT) and conserved motifs of telomerase RNA (hTR), and an H/ACA ribonucleoprotein (RNP). In the catalytic core, RNA encircles TERT, adopting a well-ordered tertiary structure with surprisingly limited protein–RNA interactions. The H/ACA RNP lobe comprises two sets of heterotetrameric H/ACA proteins and one Cajal body protein, TCAB1, representing a pioneering structure of a large eukaryotic family of ribosome and spliceosome biogenesis factors. Our findings provide a structural framework for understanding human telomerase disease mutations and represent an important step towards telomerase-related clinical therapeutics.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank P. Grob, J. Fang, A. Chintangal, P. Tobias and H. Upton for technical support; X. Zhang and J. Vogan for establishing the suspension cell transfection protocol used in early stages of this work and sharing cell lines for in vivo studies and CAB-box mutant hTR; L. Kohlstaedt and QB3 mass spectrometry facility for analysis of purified telomerase; Y. Wang, J. Jiang and J. Feigon for sharing the coordinates of the Tetrahymena catalytic core and the modelled human t/PK; E. Rodina for sharing the Hansenula polymorpha TEN coordinates before publication; and H. Upton and A. Deshpande for comments on the manuscript. We thank the National Energy Research Scientific Computing Center supported by the Office of Science of the US Department of Energy for providing computing resources under contract number DE-AC02-05CH11231. This work was funded by N.I.H. grant GM054198 to K.C. T.H.D.N. is a Fellow of the University of California, Berkeley Miller Institute for Basic Research in Science. B.J.G. was supported by fellowships from the Swiss National Science Foundation (projects P300PA_160983, P300PA_174355). E.N. is a Howard Hughes Medical Investigator.Reviewer information
Nature thanks S. Scheres and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Extended data figures and tables
Extended Data Fig. 1 Protein identification by immunoblotting, enriching active telomerase, substrate pre-binding, and comparison of intact, ΔTCAB1, and TERT–hTRmin RNPs.
a, Immunoblotting of TERT, TCAB1, dyskerin, GAR1, NHP2 and NOP10 in telomerase purified after CHAPS lysis protocol as shown in Fig. 1b. We used primary antibodies against each protein, except ZZ-SS-TERT, for which we used rabbit IgG. Owing to the wide range of molecular weights of the proteins in our sample, TERT, TCAB1, dyskerin and GAR1 were detected in one blot, while NHP2 and NOP10 were detected in a separate blot. The use of the same sample to probe all proteins was performed only once, but TERT, dyskerin and TCAB1 were also probed individually twice. b, Silver-stained SDS–PAGE gel of purified telomerase fractions obtained from adherent cells lysed using the hypotonic lysis method, which enriches active telomerase. This experiment was repeated more than five times with similar results. c, Direct primer-extension assays of the purified telomerase fractions shown in b, confirming that E1 is no longer inactive (left), and of the substrate-bound purified telomerase fractions with additional DNA substrate omitted from the assays (right). The activity observed confirmed that purified telomerase contains the DNA substrate. The activity assays with substrate added were repeated over five times and the activity assays with substrate pre-bound were repeated twice. All repeats showed similar results. d, Silver-stained SDS–PAGE gel of purified intact and ΔTCAB1 telomerase and TERT–hTRmin telomerase prepared for subunit assignments. This experiment was done only once to provide a direct comparison between these different purified telomerase complexes. e–g, Negative-stained 2D class averages of intact and ΔTCAB1 telomerase and TERT–hTRmin, respectively. h, Comparison of representative 2D class averages of intact and ΔTCAB1 telomerase and TERT–hTRmin showing the inferred localization of TCAB1 and TERT. For gel source data, see Supplementary Fig. 1.
a, Western blot detection of ZZ-SS-TERT in TERT knockout (KO) cells rescued by untagged TERT or ZZ-SS-TERT expression. Whole-cell extracts were probed using Strep antibody. HCT116 is the parental cell line. Lysate prepared from HEK 293T cells transiently transfected with ZZ-SS-TERT and hTR was used as a positive control (Ctrl). Tubulin was detected as a loading control. This experiment was performed only once to confirm the success of ZZ-SS-TERT incorporation into the HCT116 TERT KO cells. b, Telomeric restriction fragment analysis of HCT116 parental cells, TERT KO cells (before senescence), and TERT KO cells rescued with untagged or ZZ-SS-TERT transgene. Transgene-expressing cells were sampled at 31, 62 and 98 days post-transfection with transgene vectors. This experiment was performed twice with similar results. c, TRAP assay detection of telomerase activity in HCT116 parental cells, TERT KO cells, and TERT KO cells rescued with untagged TERT or ZZ-SS-TERT transgene. Whole-cell extracts were normalized by total protein concentration and assayed at 100, 30 or 10 ng of total protein per reaction. IC, internal control. This experiment was repeated four times with similar results. d, Quantification of Q-TRAP assay detection of telomerase activity in HCT parental cells, TERT KO cells and TERT KO cells rescued with untagged TERT or ZZ-SS-TERT transgene. Error bars were calculated by taking the s.d. of the average ΔCt from four time points. Data points were shown as overlays. e, Direct primer-extension assay of telomerase after template-complementary oligonucleotide purification from extracts of TERT KO cells rescued by untagged TERT or ZZ-SS-TERT transgene. Assays were performed on clarified cell lysate (crude), flow-through (O-FT) and elution (OE) using equivalent amounts of cell extract. This experiment was performed only once to re-confirm the results in c. For gel source data, see Supplementary Fig. 1.
a, Representative raw micrograph. We collected a total of 11,654 micrographs for this study. b, Representative 2D class averages obtained from reference-free 2D classification. c, Data processing strategy used in this study.
a, Representative 2D class averages obtained from 2D classification without alignment of particles that were aligned on either the catalytic core or the H/ACA lobe. For both cases, the other lobe adopts a wide range of conformations, as illustrated by the blurriness of the density. b, FSC curves for the overall map and the maps of the catalytic core and H/ACA lobe resulting from focused classification with signal subtraction and gold-standard refinement. c, Model versus map FSC curves for the catalytic core and the H/ACA RNP. We fitted only homology models as rigid bodies into the map and did not perform model coordinate refinement owing to the limited resolutions of the maps. Therefore, we used a lower FSC threshold of 0.25 for resolution estimates. d, e, Local resolution for the catalytic lobe (d) and the H/ACA lobe (e) estimated by RELION 2.064. Most of the catalytic core is resolved at 6–8 Å while most of the H/ACA lobe is resolved at 7–9 Å. f, Front (left) and back (right) views of the reconstruction showing modelled (grey) and unmodelled (gold) density. Most of the unmodelled density corresponds to single-stranded RNA regions or RNA bulges, and human protein extensions that cannot be built de novo at this resolution.
a–d, Domains of TERT. a, The TEN domain from Tetrahymena34 (PDB 2B2A). b, The truncated medaka TRBD domain33 (PDB 4O26). c, d, The RT and CTE domains from Tribolium13 (PDB 3KYL). e, Front (top) and back (bottom) views of the 5′ hairpin set of H/ACA proteins (dyskerin, red; GAR1, cyan; NOP10, wheat; NHP2, pink) bound to P4 stem (dark blue) fit by the archaeal H/ACA RNP24 (PDB 2HVY). f, Front (top) and back (bottom) views of the 3′ hairpin set of H/ACA proteins using the same model and colour scheme as e. g, Homology model of TCAB1 WD40 domain. h, Front (top) and bottom (bottom) views of hTR in the catalytic core. i, hTR in the H/ACA lobe.
Extended Data Fig. 6 Sequence alignment of TERT with secondary structure assignments based on known structures.
a, Sequence alignment of Tetrahymena and human TEN domains. The secondary structure assignments of the Tetrahymena TEN domain34 (PDB 2B2A) are shown above the aligned sequences. Regions removed before fitting are indicated with dashed lines below the sequences. b, Sequence alignment of the Tribolium, human and Tetrahymena TERT, with the latter two N-terminally truncated to match Tribolium. Secondary structure assignments of the Tribolium TERT are shown on top, with conserved motifs labelled in blue. Throughout the figure, the η symbol refers to a 310-helix. Strict β-turns and strict α-turns are displayed as TT and TTT. The three catalytic aspartic acids are indicated with black arrowheads. ESpript was used to generate this figure76.
Extended Data Fig. 7 Selected protein–protein and protein–RNA interactions in telomerase holoenzyme and comparisons between human and Tetrahymena TERT.
a, Interactions between the RT and CTE domains of TERT and the substrate–template duplex. The RT domain is divided into two subdomains, the palm (green) and fingers (orange), that are commonly observed in retroviral reverse transcriptases. The CTE (cyan) is the putative thumb. The IFD insertion that is missing in the Tribolium TERT is indicated. b, Region of the cryo-EM reconstruction shown in a. Unassigned density close to the IFD insertion is highlighted in magenta. c, Cryo-EM density of the TEN domain in the same view as that in Fig. 4b. Connecting density is observed between the template region and the P2a.1 stem. d, Map of the CR4/5 three-way junction (wheat) and the nearby TERT domains highlighting the position of the P6.1 loop near the interface of the CTE (cyan) and TRBD (blue) domains of TERT. This loop was not ordered in medaka CR4/5 bound to the TRBD alone33. e, Comparison of the Tribolium (left) and medaka (right) TRBD with the medaka CR4/5 domain of hTR13,33. Extensions of the medaka TRBD that did not fit the map were truncated for visualization. f, Cryo-EM map with H/ACA components fitted. g, h, Detailed views of regions boxed in f show TCAB1 interactions with dyskerin, GAR1 and the P8 stem-loop (g), and interactions between the two dyskerin molecules (h), where a cluster of DC mutations are found (Fig. 5d). i, Comparison of the human and Tetrahymena TERT superposed on the RT domain. Domains of human TERT are coloured as in Fig. 1a, while Tetrahymena TERT is coloured grey. The bound human and Tetrahymena templates are coloured dark and light red, respectively. j, Comparison of human and Tetrahymena19 catalytic cores fitted into the corresponding cryo-EM maps. Domains of TERT were coloured as in Fig. 1a and TER is coloured yellow. We used the catalytic core and H/ACA lobe densities resulting from our focused classification/refinement for the human telomerase and the overall 9.4 Å Tetrahymena telomerase map (EMD-6442).
Extended Data Fig. 8 Sequence alignments of H/ACA proteins with secondary structure assignments based on known structures.
a–d, Sequence alignments of Pyrococcus furiosus (archaeal) and human Cbf5/dyskerin (a), GAR1 (b), NOP10 (c), and L7Ae/NHP2 (d). Secondary structure assignments displayed on the top are from the archaeal H/ACA RNP structure24 (PDB 2HVY). The η symbol refers to a 310-helix. Strict β-turns and strict α-turns are displayed as TT and TTT, respectively. Known human dyskeratosis congenita and Hoyeraal–Hreidarsson disease mutations50 in H/ACA proteins are indicated with arrowheads. Blue arrowheads indicate residues that can be mapped onto the archaeal structure and black arrowheads indicate residues that were not mapped. ESpript was used to generate this figure76.
This file contains Supplementary Figure 1, which shows source images for all data obtained by gel electrophoresis in indicated figures.
This file contains Pymol session containing all models used for fitting in this study
The video first shows the cryo-EM densities of the H/ACA lobe and the catalytic core of the human telomerase at 8.2 Å and 7.7 Å, respectively, and fitting of the protein and RNA subunits into the cryo-EM densities. It is followed by the close-up views of the catalytic core and the H/ACA lobe highlighting how the subunits are assembled in our human telomerase structure