Telomerase RNA structural heterogeneity in living human cells detected by DMS-MaPseq

Abstract

Biogenesis of human telomerase requires its RNA subunit (hTR) to fold into a multi-domain architecture that includes the template-pseudoknot (t/PK) and the three-way junction (CR4/5). These hTR domains bind the telomerase reverse transcriptase (hTERT) protein and are essential for telomerase activity. Here, we probe hTR structure in living cells using dimethyl sulfate mutational profiling with sequencing (DMS-MaPseq) and ensemble deconvolution analysis. Approximately 15% of the steady state population of hTR has a CR4/5 conformation lacking features required for hTERT binding. The proportion of hTR CR4/5 folded into the primary functional conformation is independent of hTERT expression levels. Mutations that stabilize the alternative CR4/5 conformation are detrimental to telomerase assembly and activity. Moreover, the alternative CR4/5 conformation is not found in purified telomerase RNP complexes, supporting the hypothesis that only the primary CR4/5 conformer is active. We propose that this misfolded portion of the cellular hTR pool is either slowly refolded or degraded, suggesting that kinetic RNA folding traps studied in vitro may also hinder ribonucleoprotein assembly in vivo.

Fig. 1. Overview of human telomerase components and hTR structure.

a The conserved domain organization of TERT includes the telomerase essential amino-terminal domain (TEN), RNA binding domain (TRBD), reverse transcriptase domain (RT), and carboxy-terminal extension (CTE). A H2A/B dimer comprises one H2A and one H2B protein. b The vertebrate-conserved architecture of hTR includes the template/pseudoknot (t/PK, template sequence shown as red bar), CR4/5, and scaRNA domains. P, paired region. c The hTR template/pseudoknot domain (in blue) wraps around TERT and forms the pseudoknot helix P3. TRBD, telomerase RNA binding domain. d CR4/5 (in green) adopts an ‘L’ shaped three-way junction and sandwiches TERT TRBD between the P6.1 and P6a/b stems. A dimer of H2A/B binds P5 and P6.1. Structure figures made using PDB 7BG9.

Fig. 2. Population average DMS reactivity of the hTR t/PK domain.

a Normalized DMS reactivity of hTR t/PK domain (nt. 22–204, n = 4 biological replicates). Intensity of DMS reactivity colored according to the provided legend. Data are presented as means ± SD, open circles representing individual replicate values. Blue arcs designate the base pairing pattern of the canonical t/PK conformation seen by cryo-EM of assembled telomerase. b Secondary structure of the canonical t/PK conformation with DMS reactivity of the hTR population average overlaid onto the nucleotides. Stem elements (P), joining regions (J) and the template sequence are labeled. c DMS reactivity overlaid onto the cryo-EM model of assembled telomerase. PDB 7BG9. Source data are provided as a Source Data file.

Fig. 3. Population average DMS reactivity of the hTR CR4/5 domain.

a Normalized DMS reactivity of hTR CR4/5 domain (nt. 235–336, n = 4 biological replicates). Intensity of DMS reactivity colored according to the provided legend. Data are presented as means ± SD, open circles representing individual replicate values. Blue arcs designate the base pairing pattern of the canonical t/PK conformation seen by cryo-EM of assembled telomerase. b Secondary structure of the canonical CR4/5 conformation with DMS reactivity of the hTR population average overlaid onto the nucleotides. c DMS reactivity overlaid onto the cryo-EM model of assembled telomerase. PDB 7BG9. Source data are provided as a Source Data file.

Fig. 4. Evaluating data-driven RNA structure models with the area under the receiver operating characteristic curve (AUROC).

a Bar plot depicting AUROC values for DMS reactivities derived from five biological replicates of the t/PK domain (left, blue bar) and four biological replicates of the CR4/5 domain (right, green bar). Data are presented as means ± SD. b Receiver operating characteristic (ROC) curve comparing the t/PK Chen model with our DMS reactivities of the t/PK domain from four replicates (different colored lines). c ROC curve comparing the CR4/5 Chen model without DMS reactivities of the CR4/5 domain from four biological replicates (different colored lines). Source data are provided as a Source Data file.

Fig. 5. DREEM-deconvoluted DMS profiles and structure predictions of hTR t/PK.

a Normalized DMS reactivity profiles of the two clusters predicted by DREEM. Intensity of DMS reactivity colored according to the provided legend. Blue arcs designate the base pairing pattern of the data-guided predicted t/PK secondary structure. b Data-guided secondary structure prediction of the t/PK domain from the canonical cluster of DMS reactivities. c Data-guided secondary structure prediction of the t/PK domain from the alternative cluster of DMS reactivities. Source data are provided as a Source Data file.

Fig. 6. DREEM-deconvoluted DMS profiles and structure predictions of hTR CR4/5.

a Normalized DMS reactivity profiles of the two clusters predicted by DREEM. Intensity of DMS reactivity colored according to the provided legend. Blue arcs designate the base pairing pattern of the data-guided predicted CR4/5 secondary structure. b Data-guided secondary structure prediction of the CR4/5 domain from the canonical cluster of DMS reactivities. c Data-guided secondary structure prediction of the CR4/5 domain from the alternative cluster of DMS reactivities. Source data are provided as a Source Data file.

Fig. 7. Design and MaPseq validation of CR4/5 mutants.

a Canonical CR4/5 secondary structure with mutated nucleotides labeled. b Cryo-EM structure (PDB 7BG9) of CR4/5 with mutated nucleotides highlighted. c Normalized DMS reactivity profiles of clusters predicted by DREEM for overexpressed WT and CR4/5 mutants. Source data are provided as a Source Data file.

Fig. 8. Telomerase activity and RNP assembly for hTR mutants.

a Activity assay of FLAG-purified HEK293T-derived telomerase measuring incorporation of ³²P-dGTP into a telomeric DNA primer for the WT, M1, M2, and M3 hTR variant constructs. LC, labeled oligonucleotide loading controls. Experiment was repeated four times (n = 4) with similar results. b Quantification of telomerase activity. HEK293T- or HeLa-derived telomerase activity calculated by summing the total lane signal and correcting for differences in intensity of the loading controls. c Bar plot of fractional activity relative to WT was calculated from the slopes of the linear regression fits. d Western blots for hTERT (top), DKC1 (middle), and Gar1 (bottom) all performed after IP of FLAG-hTERT. Relative signal compared to wild type for two different loading amounts is shown below each lane. (Gar1 values are approximate due to accidental truncation of tops of bands.) Experiment was conducted two times (n = 2) with similar results. e Top, northern blot for hTR in HEK293T cell extracts, each hTR intensity normalized to U1 and U2 snRNA recovery controls and then to the WT hTR signal. Bottom, northern blot for hTR in anti-hTERT IP fractions, values normalized to WT. hTR migrates as a doublet when samples are heated and snap-cooled (right lanes) because the 7 M urea gel is not completely denaturing. Experiment was conducted four times (n = 4) with similar results.

Fig. 9. DREEM-deconvoluted DMS profiles and structure predictions of hTR CR4/5 from telomerase RNP complexes assembled in cells and purified via the FLAG-tag on hTERT.

a Normalized DMS reactivity profiles of the two clusters predicted by DREEM. Intensity of DMS reactivity colored according to the provided legend. Blue arcs designate the base pairing pattern of the data-guided predicted CR4/5 secondary structure. b Data-guided secondary structure prediction of the two clusters with overlaid DMS reactivities. The major and minor clusters have identical structures. Intensity of DMS reactivity colored according to the provided legend.

Fig. 10. Model of telomerase RNP assembly with respect to hTR folding.

hTR exists as a structural ensemble in cells, with a minority population adopting alternative conformations in the t/PK and CR4/5 domains (red). The canonical fold of hTR may be favored by the presence of RNA folding chaperones. hTR molecules adopting the canonical fold are efficiently co-assembled into a functional RNP, whereas hTR molecules in the alternative conformation(s) are not efficiently assembled in the RNP complex or are slowly degraded.