Structure of human telomerase holoenzyme with bound telomeric DNA

Abstract

Telomerase adds telomeric repeats at chromosome ends to compensate for the telomere loss that is caused by incomplete genome end replication¹. In humans, telomerase is upregulated during embryogenesis and in cancers, and mutations that compromise the function of telomerase result in disease². A previous structure of human telomerase at a resolution of 8 Å revealed a vertebrate-specific composition and architecture³, comprising a catalytic core that is flexibly tethered to an H and ACA (hereafter, H/ACA) box ribonucleoprotein (RNP) lobe by telomerase RNA. High-resolution structural information is necessary to develop treatments that can effectively modulate telomerase activity as a therapeutic approach against cancers and disease. Here we used cryo-electron microscopy to determine the structure of human telomerase holoenzyme bound to telomeric DNA at sub-4 Å resolution, which reveals crucial DNA- and RNA-binding interfaces in the active site of telomerase as well as the locations of mutations that alter telomerase activity. We identified a histone H2A-H2B dimer within the holoenzyme that was bound to an essential telomerase RNA motif, which suggests a role for histones in the folding and function of telomerase RNA. Furthermore, this structure of a eukaryotic H/ACA RNP reveals the molecular recognition of conserved RNA and protein motifs, as well as interactions that are crucial for understanding the molecular pathology of many mutations that cause disease. Our findings provide the structural details of the assembly and active site of human telomerase, which paves the way for the development of therapeutic agents that target this enzyme.

Extended Data Figure 1. Image processing scheme.

Summary of the data processing strategies that yielded the reconstructions described in this study.

Extended Data Figure 2. Local resolution and resolution estimation.

a, Gold-standard FSC curves for maps of H/ACA lobe, the catalytic core and the best two full-length (FL) classes. Resolutions were estimated at FSC=0.143. b, FSC curves for the refined H/ACA and catalytic core models vs the corresponding maps. Resolutions were estimated at FSC=0.5. Local resolution estimated by RELION 3.1 (ref. 54) for c, the H/ACA lobe; d, the catalytic core; e,FL-3D class 2; f, FL-3D class 5. For direct comparisons, the same local resolution range (3.3-16 Å) was used for all maps.

Extended Data Figure 3. EM density of protein components.

Full densities of a, TERT; e, histone H2A; f, histone H2B; i, TCAB1; k, 5’ H/ACA tetramer; l, 3’ H/ACA tetramer. b, Close-up view of the active site density of TERT with an empty nucleotide binding pocket (see also Extended Data Fig. 6f). c, d, Representative EM densities of TERT. g, h, j, Representative EM densities of histone H2A, histone H2B and TCAB1, respectively. m, Close-up view of the density of N-terminal extension of the 5’-dyskerin bound to a hydrophobic pocket within the 3’-dyskerin (see also Fig. 4c). n, Close-up view of the density of helix 352-375 of the 3’-dyskerin bound to the equivalent hydrophobic pocket as in panel (m) within the 5’-dyskerin (see also Fig. 4d). o, Close-up view of the density of hTR P8 stem loop containing the CAB- and BIO-boxes, which interact TCAB1 and the 3’ NHP2 (see also Extended Data Figs. 4l and 8d). p, Close-up view of the density of the H- and ACA-boxes interacting with the two dyskerin molecules (see also Extended Data Figs. 4k and 8c).

Extended Data Figure 4. EM density of hTR and DNA substrate.

a, Secondary structure schematic of hTR based on the structure. This figure was modified from the telomerase database. b, c, Full density of hTR in the catalytic core and the H/ACA lobe, respectively. d, Close-up view of the density of DNA substrate-hTR template interactions (see also Fig. 2b). e, Close-up view of density of the DNA substrate and neighboring TERT residues (see also Extended Data Fig. 6d). f, Close-up view of density of the RNA template region and neighboring TERT residues (see also Extended Data Fig. 6e). g, Density of the P6.1 hairpin of the CR4/5 domain. Labelled residues are highlighted in Fig. 3a. h, Close-up review of the density of the P6.1 stem loop interacting with residues of the CTE domain of TERT (see also Fig. 3a and panel (g)). i, Close-up view of the density of residue L1019 of TERT, which interacts with the two flipped-out nucleotides, U177 of the PK and U307 of the P6.1 stem loop as highlighted in Fig. 3b. j, Representative density of the PK containing the base-triples, which are highlighted in black. Nucleotide U113 is modelled but not visible in this view. k, Density of the H- and ACA-boxes (see also Extended Data Fig. 3p for a related view). l, Density of the P8 stem loop (see also Extended Data Fig. 3o for a related view).

Extended Data Figure 5. Multibody refinement and conformational dynamics analysis of the full-length structure.

a, Summary of the multibody refinement strategy and principal component analysis. The best two subsets from global 3D classification were subjected to multibody refinement using two masks for the H/ACA lobe (yellow) and the catalytic core (cyan). b, Principal component analysis for 3D class 2. c, Principal component analysis for 3D class 5. The first and last frames of the eigenvector series of the first 6 principal components are shown. Curved arrows indicate the movements. d, Top 10 hTR ensembles modelled by DRRAFTER into the refined FL 3D class 2 map (see Supplementary Data 2). e, Top 10 hTR ensembles modelled by DRRAFTER into the refined FL 3D class 5 map (see Supplementary Data 3). hTR is shown in blue, and the protein subunits are in grey.

Extended Data Figure 6. Telomerase catalytic cycle and DNA path.

a, Domain architecture of human TERT and conserved motifs commonly observed in reverse transcriptases. b, TERT conserved motifs shown in (a) and their interactions with the template and DNA substrate. c, A model for the telomerase catalytic cycle. The template region of hTR is divided into an alignment region and a templated region. The telomeric DNA repeat first binds to the alignment region, followed by 6 consecutive nucleotide additions using the templated region. After the synthesis of the full telomeric repeat, the DNA substrate translocates to bind the alignment region to start another round of repeat synthesis. The state captured in our structure is indicated with an asterisk. d, Interactions between the 3’ telomeric TTAGGG repeat of the substrate and TERT. e, Interactions between the template region of hTR and TERT. f, TERT active site in a pre-nucleotide state. D712, D868 and D869 form the catalytic triad for nucleotide addition. See also Extended Data Fig. 3b. g, Modelled dTTP (PDB 1T3N) in the vacant nucleotide site of TERT. The C2 ribose is indicated. h, Structure of human TERT with PK/t and CR4/5 domains of hTR and DNA. i, Electrostatic surface potential of human TERT with hTR and the DNA substrate shown in the same view as (h). The highlighted human TEN-IFD-TRAP interface (in blue) is positively charged and could potentially bind the 5’ end of the DNA substrate in human telomerase. j, IFD-TRAP and TEN domains of TERT with residues known to affect TPP1 binding to TERT highlighted as spheres. The proposed DNA path would bring it close to the proposed TPP1 binding site on the TEN domain. k, Model of TPP1-POT1 bound to human TEN domain based on the Tetrahymena p50-TEB complex. Despite the similar overall domain arrangements, the domains of Tetrahymena and human TERT do not align well as whole. To obtain the model, we superposed the Tetrahymena TEN domain-p50-TEB1-2-3 complex (PDB 6D6V) onto the human TEN domain. l, Model of telomerase catalytic cycle. Telomerase template RNA binds the telomeric DNA substrate by base-pairing. The DNA binding sites on TERT are indicated with the yellow stars. One binding site is provided by motif T and the CTE domain of TERT near the active site, as observed in the structure (Fig. 2d and panel (d)). The second binding site is proposed to be provided by the TEN domain at the 5’ end of the DNA. After the synthesis of a full telomeric repeat, the nascent DNA undergoes translocation and rebinding to the template RNA. We propose that the two DNA binding sites form an anchor site to allow DNA retention for multiple rounds of repeat synthesis.

Extended Data Figure 7. Identification of H2A-H2B as human telomerase holoenzyme subunits.

a, 3D classification of the catalytic core (dataset 1) showing the presence of the unaccounted density (in yellow). The best class (boxed) has the most homogenous density and was selected for the final refinement. Similar observations were made with the second dataset. b, The 8 Å catalytic core map (in grey) with the previously unmodelled density (in yellow) (left panel) and with the model of the catalytic core obtained from this work fitted into it (right panel). The density assigned as the histone H2A-H2B dimer coincided with the unmodelled density from the previous work. c, The refined catalytic core map with hTR, TERT and H2A-H2B segmented in different colors. d, Interactions between CR4/5 and the H2A-H2B dimer in human telomerase. The bottom panel shows the electrostatic surface potential of the histone dimer and the positively charged surface used for interacting with the CR4/5. e, Nucleosome structure with the H2A-H2B dimer colored and oriented the same way as shown in d. This shows that the H2A-H2B uses the same surface to bind both nucleosomal DNA and CR4/5 (PDB ID 1KX5 (ref. 27)). f, Purified H2A-H2B and CR4/5 RNA used for EMSA in Fig. 3f. No RNA ladders were loaded with the CR4/5 RNA. g, Immunoblots of crude lysate of 293T cells transfected with Twin-Strep TERT and hTR expression constructs (input) and O-elution from 2’-O-Methyl purification (Fig. 3c). These samples were immunoblotted for Strep, H2A, H2B, H3 and H4. The presence of H2A, H2B, H3 and H4 is also confirmed by mass spectrometry (Supplementary Data 4). h, Immunoblots of crude lysate of 293T cells (input) and O-elution from 2’-O-Methyl purification. These samples were immunoblotted for dyskerin, H2A, H2B, H3 and H4. i, Structure of the Tetrahymena TRBD-CTE-p65 and stem loop 4 (SL4) (PDB 6D6V). j, Structure of human TRBD-CTE-H2A-H2B and CR4/5. k, Quantification of the electrophoretic mobility shift assays shown in Fig. 3f for K _d determination. ─Total Fraction Bound∥ reflects quantification of free probe depletion against total probe with increasing histone concentration. ─Specific Fraction Bound∥ reflects quantification of increasing discrete shifted complex band against total probe with increasing histone concentration. Points represent values from three independent replicates (Supplementary Fig. 1). h, Superposition of the histone octamer structure, with flexible histone tails removed, onto the H2A-H2B dimer bound to the human telomerase catalytic core in two different views (PDB 1KX5 (ref. 27)).

Extended Data Fig. 8. H/ACA RNP and molecular interactions of conserved RNA motifs.

a, Front (right) and back (left) views of the H/ACA RNP with subunits colored as indicated. b, Secondary structure schematic of the H/ACA domain of hTR. c, Close-up view (left panel) of the H- and ACA-boxes and their interactions with each other and with the two dyskerin molecules (see also Extended Data Figs. 3p, 4k). The right panel shows their sequences with conserved nucleotides highlighted in bold. d, Close-up view (left panel) of P8 stem loop and interactions between the CAB- and BIO-boxes with TCAB1 and NHP2 (see also Extended Data Figs. 3o, 4l). A schematic diagram of these interactions is shown in the right panel.

Figure 1. Cryo-EM structure of human telomerase holoenzyme.

a, Front (left) and back (right) views of the composite cryo-EM maps of the catalytic core and the H/ACA lobe at 3.8 Å and 3.4 Å, respectively, with subunits colored as indicated. b, Front view of the structure with hTR and the DNA substrate highlighted. c, Domain architectures of protein subunits. TEN, telomerase essential N-terminal domain; TRBD, telomerase RNA-binding domain; RT, reverse transcriptase; IFD-TRAP, insertion in fingers-TRAP domain; CTE, C-terminal extension; NTE, N-terminal extension; TruB, tRNA pseudouridine synthase B-like domain; PUA, pseudouridine synthase and archaeosine transglycosylase; WD40, Trp-Asp 40 repeat domain; RBD, RNA-binding domain. d, Secondary structure of hTR and DNA substrate (see also Extended Data Fig. 4a). TEM, template; PK/t, pseudoknot/template; CR4/5, conserved regions 4 and 5; BIO, biogenesis-promoting.

Figure 2. Human telomerase catalytic core.

a, Domain architecture of TERT. This domain color scheme is used throughout the paper unless otherwise stated. b, TERT structure with the PK/t domain of hTR and DNA substrate. The inset shows a close-up view of the DNA substrate-RNA template duplex in TERT active site (see also Extended Data Fig. 4d) c, Cartoon of TERT domains, hTR PK/t domain and the DNA substrate. d, Schematic diagram of the DNA-RNA duplex observed in our structure and interactions with TERT residues (see also Extended Data Figs. 6d-f).

Figure 3. CR4/5 interactions with TERT and novel H2A-H2B subunits.

a, CR4/5 domain of hTR and its interaction with TERT TRBD and CTE domains and the histone H2A-H2B dimer. The inset provides a close-up view to highlight the recognition of the P6.1 apical loop by TERT and H2A (see also Extended Data Figs. 4g, h). b, The close proximity between the PK (orange) and P6.1 stem loop (wheat) (see also Extended Data Fig. 4i). c, Schematic of 2’-O-Methyl oligo-purification coupled with immunoprecipitation (IP) to confirm the presence of H2A-H2B in human telomerase structure. SS, twin strep tag. d, Activity assay results of the IP shown in panel (c). Crude, input lysate; O-FT, flow-through from oligo-purification; OE, elution from oligo-purification; FT, flow-through from immunoprecipitation with each set of antibodies; B, bound sample on beads from immunoprecipitation with each set of antibodies; 12nt RC, 12-nucleotide recovery control; 18nt, 18-nucleotide marker. We performed two separate quantifications. Each FT activity was quantified relative to input (OE) activity. Each bound activity was quantified relative to anti-Strep IP positive control. All quantifications were normalised to the signal of the recovery control. e, Immunoblot analyses of the IP shown in panel (c). Each fraction was immunoblotted with NHP2 antibody (α-NHP2) and NOP10 antibody (α-NOP10). Signal from each flow though fraction (FT) was quantified relative to the input (OE) signal. Signal from each bound fraction (Bound) was quantified relative to the anti-Strep IP signal. f, Native electrophoretic mobility shift assays showing titration of purified H2A-H2B against P end-labelled CR4/5 RNA. See Extended Data Fig. 7k for quantification. Experiments shown in (d)-(f) were done in three technical replicates (see also Supplementary Figs 1 and 2).

Figure 4. H/ACA RNP structure and human disease mutations.

a, Schematic of the H/ACA RNP showing hTR interactions with the H/ACA proteins. b, Overall structure of the H/ACA RNP with DC and HH disease mutations highlighted in spheres. Close-up view of the disease mutation hotspot, which lies at the interface between the two dyskerin molecules (inset). c, The hydrophobic pocket of 3’ dyskerin, which accommodates the N-terminal extension of the 5’ dyskerin (see also Extended Data Fig. 3m). d, The hydrophobic pocket of 5’ dyskerin, which accommodates helix 352-357 of 3’ dyskerin (see also Extended Data Fig. 3n).