Cryo-EM structure of human telomerase dimer reveals H/ACA RNP-mediated dimerization

Abstract

Telomerase ribonucleoprotein (RNP) synthesizes telomeric repeats at chromosome ends using a telomerase reverse transcriptase (TERT) and a telomerase RNA (hTR in humans). Previous structural work showed that human telomerase is typically monomeric, containing a single copy of TERT and hTR. Evidence for dimeric complexes exists, although the composition, high-resolution structure, and function remain elusive. Here, we report the cryo-electron microscopy (cryo-EM) structure of a human telomerase dimer bound to telomeric DNA. The structure reveals a 26-subunit assembly and a dimerization interface mediated by the Hinge and ACA box (H/ACA) RNP of telomerase. Premature aging disease mutations map to this interface. Disrupting dimer formation affects RNP assembly, bulk telomerase activity, and telomere maintenance in cells. Our findings address a long-standing enigma surrounding the telomerase dimer and suggest a role for the dimer in telomerase assembly.

Fig. 1. Structure of the human telomerase dimer.

(A) Schematic showing secondary structure of hTR. (B) Schematic of human telomerase monomer. The domains of hTR are colored as shown in (A). (C) Representative cryo-EM 2D class averages of the telomerase monomer and the telomerase dimer. (D) Schematic of the purification strategy to probe for the human telomerase dimer using ZZ-SS-TERT and ZZ-F-TERT. ZZ, protein A; SS, twin-Strep tag; F, Flag tag; OE, oligonucleotide elution; Strep FT, Strep-Tactin flow-through; Strep E, Strep-Tactin elution. (E) 6.2 Å consensus cryo-EM map of the full human telomerase dimer. The two protomers are colored in magenta and blue. (F) The consensus cryo-EM map of the full telomerase dimer with fitted two copies of the 3.0 Å cryo-EM reconstruction of the human telomerase H/ACA RNP protomer and the 3.3 Å cryo-EM reconstruction of the human telomerase catalytic core. These composite maps were obtained from focused classification and refinement on the H/ACA RNP and the catalytic cores separately. (G) Atomic model of the full telomerase dimer (left) as shown in (F) and the schematic of the structure (right) as shown in (F). The model was obtained through a combination of fitting published models, manual building and DRRAFTER modeling (23, 26, 27). (H) Immunoblots of the telomerase sample with ZZ-SS-TERT and ZZ-F-TERT (left), and control telomerase with ZZ-F-TERT (right) obtained by purification strategy depicted in (D). An anti-Flag antibody was used.

Fig. 2. The human telomerase dimer is catalytically active.

(A) Telomerase activity assay of the purified dual-tagged human telomerase dimer following the procedure depicted in Fig. S5A. RC, recovery control. (B) Domain organization of human TERT (hTERT), histones H2A and H2B, and TPP1. TEN, telomerase essential N-terminal; PAL, proline/arginine/glycine-rich linker; TRBD, telomerase RNA binding domain; RT reverse transcriptase; IFD, insertion in the fingers domain; CTE, C-terminal extension; OB, oligonucleotide/oligosaccharide binding; PBM, POT1 binding motif; TBM, TIN2 binding motif. (C) 3.3 Å cryo-EM map of the human telomerase catalytic core using C2-expanded particles. Domains are labelled and colored according to the color scheme in (B). (D) The consensus cryo-EM map of the full human telomerase dimer with fitted 3.8 Å and 3.6 Å cryo-EM reconstructions of the human telomerase catalytic core 1 and 2, respectively. The maps were generated without applying symmetry functions. Insets show the close-up views of the DNA-bound telomerase active site of the catalytic core 1 (left) and catalytic core 2 (right). (E) Data processing strategy used to obtain the human telomerase catalytic cores from the shared particles, which originated from the subsets that gave the reconstructions of the two catalytic cores shown in (D). The shared particles are those found in both particle stacks used to generate catalytic core 1 and catalytic core 2 maps. The maps of telomerase catalytic cores 1 and 2, obtained from only the shared particles, are shown with the DNA in the active sites.

Fig. 3. Dimerization relies on interactions between the two H/ACA RNPs.

(A) 3.9 Å cryo-EM map of the H/ACA RNP dimer with individual protomers colored. Only subunits of protomer 2 (blue) and hTR of protomer 1 are labelled for simplicity. (B, C) A cartoon representation of the telomerase H/ACA RNP dimer model. In (C), only protomer 2 and the P4/5 linker of protomer 1 are highlighted in colors. (D) Schematic of the H/ACA RNP dimer. (E) A close-up view of the inter-protomer interactions between the hTR of protomer 1 and the 5’ H/ACA heterotetramer of protomer 2. The P4/5 linker interacts with the 5’ dyskerin, 5’ NHP2 and 5’ NOP10. (F) A close-up view of the interactions between the hTR P4.2/P5 junction of protomer 1 and the 5’ dyskerin and 5’ NHP2 of protomer 2. (G) A close-up view of the interactions between the hTR P4.2 stem of protomer 1 and the 5’ NHP2 and 5’ NOP10 of protomer 2. (H) Schematic of the interactions between the P4/5 linker of hTR from protomer 1 and the 5’ H/ACA proteins of protomer 2.

Fig. 4. Disrupting the P4.2 stem of hTR affects the binding of dyskerin and telomere maintenance in cells.

(A) Schematic of WT hTR P4.2 stem, P4.2 switch and P4.2 compensatory (P4.2 comp.) hTR mutants. Nucleotides 329 and 330 of hTR (highlighted in with green arrows) are also mutated in experiments shown in (B) and (C). Nucleotides mutated to create the P4.2 switch and P4.2 comp. mutants are shown in red. (B) Immunoblots and northern blots of the human telomerase reconstituted with the ZZ-SS-TERT, ZZ-F-TERT, and the WT or mutant hTR. In the left panel, TERT levels in telomerase lysates (input) are normalized to tubulin levels. Telomerase was purified as described in Fig. 1D. The right panel depicts the immunoblot and corresponding quantification of the levels of the ZZ-F-TERT detected in the Strep elution (Strep E). ZZ-F-TERT levels in the Strep E of mutants are normalized to that in the WT sample. The bar graph represents the amount of ZZ-F-TERT in the Strep E assembled with the mutant hTR compared to the WT hTR. Experiments were performed in triplicates (n = 3). Error bars represent the standard error of the mean (SEM), and significant p value is reported for the P4.2 switch hTR mutant. See fig. S13 for the replicate data. (C) Telomere restriction fragment (TRF) assay of HCT116 hTR knockout (KO) cell lines transduced with either WT or mutant hTR from (A). Lanes 1 and 2 show telomere lengths of the parental HCT116 and the KO cell lines, respectively. Lanes 3 to 14 show telomere lengths of KO cells that were rescued with various hTR transgene constructs. Yellow dash lines indicate the mean telomere length for each lane. (D) Immunoblots of the input lysates (input), OE, and strep E to dissect the effects of disrupting P4.2 stem of hTR. Each of hTR constructs, either WT or P4.2 switch or P4.2 comp. mutant, was reconstituted with a mixture of ZZ-SS-TERT and ZZ-F-TERT (fig. S14A). Telomerase was purified via the O-purification and Strep-Tactin pulldown, as shown in fig. S14A. Experiments were performed in triplicates (n = 3). See fig. S14B for replicate data. (E) Bar graph showing the levels of TERT in the OE of WT hTR, P4.2 switch and P4.2 comp. mutants. (F) Bar graph showing the levels of dyskerin in the OE of WT hTR, P4.2 switch and P4.2 comp. mutants. (G) Bar graph showing the levels of dyskerin relative to TERT in the OE of WT hTR, P4.2 switch and P4.2 comp. mutants. Together with the graphs shown in (E) and (F), this graph shows that in the OE, dyskerin levels in the P4.2 switch hTR mutant were more severely reduced than TERT compared to the WT hTR. (H) Bar graph showing the levels of dyskerin in the Strep E of WT hTR, P4.2 switch and P4.2 comp. mutants. In the quantification shown in (E) to (H), TERT or dyskerin levels of the mutants were normalized to those of the WT hTR. Error bars in (E) to (H) represent the standard error of the mean (SEM), and significant p values are also reported.

Fig. 5. Dimerization plays a role in telomerase assembly.

(A) Disease mutation mapping of the H/ACA proteins and hTR onto the model of the H/ACA RNP dimer (6, 53). Disease mutations are represented as spheres. (B) Inset showing a close-up view of the interface between the hTR P4.2 stem of protomer 1 and the H/ACA proteins of protomer 2 with the disease mutations at this interface. (C) Immunoblots of cell lysates (input) and OE from cells transfected with ZZ-SS-TERT, hTR, and WT or mutant 3xF-dyskerin. The total dyskerin levels in the OE were calculated as 3xF-dyskerin plus endogenous dyskerin. Mutant 3xF-dyskerin levels over total dyskerin were normalized to the WT sample. (D) Immunoblots of cell lysates and OE from cells transfected with ZZ-SS-TERT, hTR, and either WT or mutant 3xF-NHP2. Analysis and quantification are the same as for (C). (E) Bar graph showing the levels of 3xF-dyskerin over the total amount of dyskerin in the OE samples from (C). (F) Bar graph showing the levels of 3xF-NHP2 over the total amount of NHP2 in the OE samples from (D). Experiments in (C) and (D) were performed in triplicates (n = 3). Also see fig. S17 for the replicate data. Error bars in (E) and (F) represent standard error of the mean (SEM), and significant p values are reported.

Fig. 6. GAR1 role in telomerase dimerization.

(A) Consensus cryo-EM density of the H/ACA RNP dimer with the 5’ G-quadruplex (GQ), the 3’ GAR1 from protomer 1 and the 3’ GAR1 of protomer 2 fitted into the density. (B) Interaction between the 3’ GAR1 subunits of the H/ACA RNP protomers 1 and 2 within the 3.9 Å H/ACA RNP dimer map. Although the density for the N-terminal arginine-glycine (RG) rich domains of both protomers is clear, they do not appear to have defined structures for model building. The dash lines and the green cryo-EM density represent the predicted positioning of the N-terminal RG domains of the 3’ GAR1 subunits. (C) The density of the N-terminal RG domains of the two 3’ GAR1 subunits reaching towards the hTR P7b stem of both H/ACA RNP protomers. (D) Schematic of the two P7b stems of hTR within the dimer and their contact with the N-terminal RG domains of the 3’ GAR1. (E) 3.9 Å cryo-EM H/ACA RNP dimer map with fitted ensemble of the 5’ GQ conformations, modelled by DRRAFTER (26). (F) Close-up view of the 5’ GQ interaction with the 5’ GAR1 within protomer 1. The same interaction is formed between the 5’ GQ and the 5’ GAR1 of protomer 2. (G) Consensus cryo-EM density of the full telomerase dimer with histone H2A-H2B dimer from protomer 1 and the 5’ GAR1 of protomer 2 fitted into the density. (H) A close-up view showing the interface between the histone H2B of protomer 1 and the 5’ GAR1 of protomer 2.