Exosome-mediated decay of unstable long extended precursors of human telomerase RNA is dependent on 5′-cap trimethylation

In this study, Samajdar et al. describe how sequential posttranscriptional modifications, particularly 5′-cap trimethylation by TGS1 and 3′-end oligoadenylation by PAPD5, synergistically coordinate the maturation and exosome-mediated decay of human telomerase RNA (hTR). Inhibiting these steps could rescue hTR levels and telomerase function in cells harboring pathogenic telomerase mutations, highlighting their therapeutic potential for telomere biology disorders.

Abstract

Mutations that impact maturation of human telomerase RNA (hTR) are common in telomere biology disorders. Here, we describe sequential posttranscriptional modifications that coordinate hTR biogenesis and decay. Initially, TGS1-mediated 5′-cap trimethylation targets long genomically extended hTR precursors for degradation. Prevention of 5′-cap trimethylation results in accumulation of nucleolar 3′-end extended precursors, evading MTR4 recognition and degradation by the exosome. In a second step, 3′-end oligoadenylation by PAPD5 promotes degradation of mature hTR, a process that remains dependent on 5′-cap modifications, as prevention of trimethylation inhibits decay of heavily 3′-end oligoadenylated molecules. Combined inhibition of 5′-cap trimethylation and 3′-end oligoadenylation synergistically increases hTR in cells harboring pathogenic mutations in telomerase. These data reveal a precise interplay between 5′- and 3′-end posttranscriptional modifications that dictate hTR fate and highlight the potential of RNA therapeutics for treatment of telomere biology disorders.

Noncoding RNAs are composed of different RNA species with various physiological and pathological functions (Cable et al. 2021). One such species is represented by long noncoding RNAs (lncRNAs), RNA transcripts of >500 nt that are mostly generated by RNA polymerase II (Mattick et al. 2023). lncRNAs are implicated in the pathology of different diseases, including hematopoietic failure, cardiovascular disease, and cancer, and their modulation represents a novel alternative for therapeutic approaches (Adams et al. 2017). The human telomerase RNA (hTR) component is an example, as impaired hTR biogenesis, maturation, and function are responsible for most cases of telomere biology disorders (TBDs), a group of diseases precipitated by telomere dysfunction. While TBDs show myriad phenotypes, they are usually characterized by severe hematopoietic failure, pulmonary fibrosis, liver failure, skin abnormalities, and cancer predisposition (Revy et al. 2023).

The assembly of the telomerase ribonucleoprotein (RNP) complex revolves around the stepwise processes that regulate the maturation of its RNA component. While the structure and size of mature telomerase RNA vary significantly (from 451 nt in humans to >1300 nt in yeast) (Theimer and Feigon 2006; Podlevsky et al. 2008), they serve as a platform for the assembly of the entire telomerase complex in addition to being a template for the reverse transcriptase function of the telomerase reverse transcriptase (TERT). Also crucial for proper telomerase function, telomerase RNA regulates telomerase accumulation and localization (Chen and Batista 2025). In humans, while most hTR molecules are found diffused around the nucleus (Schmidt et al. 2016, 2018; Laprade et al. 2020), telomerase localization to Cajal bodies (CBs) is dependent on the binding of hTR to TCAB1 (Venteicher et al. 2009; Laprade et al. 2020). Structurally, hTR contains a H/ACA domain that binds to two sets of the dyskerin complex (composed of DKC1, NOP10, NHP2, and NAF1), a reaction that happens cotranscriptionally and is essential for hTR stability (Darzacq et al. 2006). Mutations in all components of the dyskerin complex have been identified in TBD patients, where they cause reduced hTR levels, low telomerase activity, and exacerbated telomere shortening (Armanios and Blackburn 2012; Revy et al. 2023). However, despite the clear relevance of hTR for cell physiology, the biogenesis of this noncoding RNA remains incompletely understood.

hTR molecules are initially transcribed as 3′-end extended transcripts or precursors that can reach up to 1000 nt in length (Roake et al. 2019). These extended hTR precursors can be of different lengths, have reduced telomerase activity (Deng et al. 2019), and are either degraded or slowly processed into the mature, 451 nt long molecule (Tseng et al. 2018; Roake et al. 2019). While still not completely elucidated, the process of “trimming” hTR precursors is performed primarily by the 3′-to-5′ exonuclease poly(A)-specific ribonuclease (PARN), but target of Egr1 (TOE1) has also been shown to process hTR precursors into functional molecules (Tseng et al. 2018; Roake et al. 2019). Over the last few years, different posttranscriptional modifications have emerged as major regulators of hTR biogenesis, localization, and stability (Revy et al. 2023). Following transcription, hTR molecules are oligoadenylated at their 3′ ends by the noncanonical poly(A) polymerase PAPD5, which acts as a signal for the exosome-mediated degradation of hTR (Moon et al. 2015; Tseng et al. 2015; Shukla et al. 2016). Acting in opposition to PAPD5, PARN removes oligoadenylated tails from short 3′-extended hTR precursors, thereby promoting their maturation and preventing premature degradation by the exosome (Moon et al. 2015; Shukla et al. 2016; Roake et al. 2019). Highlighting the relevance of this pathway for telomerase function, mutations in PARN are found in TBD patients, causing low levels of mature hTR and leading to exacerbated telomere shortening (Dhanraj et al. 2015; Stuart et al. 2015; Tummala et al. 2015). Recruitment of the RNA exosome complex to hTR is performed by the nuclear exosome targeting (NEXT) complex, composed of the nuclear helicase MTR4, the RNA binding protein RBM7, and the zinc finger protein ZCCH8 (Lubas et al. 2011). Mutations in the NEXT component ZCCHC8 have also been found in patients with short telomeres (Gable et al. 2019), with an accumulation of nonfunctional 3′-end extended hTR precursors leading to reduced telomerase activity (Gable et al. 2019).

Less is known about the role of 5′-end modifications during hTR processing and stability. Concomitantly with its transcription, hTR is initially 7-methyl guanosine (m7G)-capped at its 5′ end (Tseng et al. 2015). Monomethylguanosine (MMG)-capped hTR molecules are then trimethylated to form N2, 2, 7 trimethylguanosine (TMG) caps by trimethylguanosine synthase 1 (TGS1). Silencing of TGS1 prevents the formation of TMG-capped hTR molecules, which increases hTR levels and telomerase activity in human cells (Chen et al. 2020; Buemi et al. 2022; Galati et al. 2022). However, the molecular mechanisms that prevent hTR decay in TGS1-depleted cells, as well as the role of the exosome RNA degradation complex in this process, remain obscure. Likewise, it remains unknown whether the 3′- and 5′-end posttranscriptional modifications to hTR act in a coordinated fashion during hTR biogenesis to stabilize precursor and mature hTR molecules in human cells. Finally, the extent to which these processes can be modulated to restore telomerase activity in cells harboring pathogenic mutations has not been explored (Batista et al. 2022).

To understand how different posttranscriptional modifications to hTR are coordinated and to determine their function throughout hTR processing, we developed systems where we can independently and in combination modify 5′-cap methylation and 3′-end oligoadenylation levels of hTR precursor and mature molecules. Analysis of different steps of hTR biogenesis and localization shows that prevention of 5′ TMG cap formation by inhibition of TGS1 leads to accumulation of long 3′-end extended hTR precursors (up to 1000 nt), a process that is exacerbated in DKC1 mutant cells, where hTR is usually quickly degraded (Revy et al. 2023). Our data indicate that these 5′-end MMG-capped precursors are localized to the nucleolus and not recognized by the nuclear MTR4 helicase, an essential cofactor of the nuclear exosome complex that targets polyadenylated nuclear RNAs for degradation (Lubas et al. 2015; Zinder and Lima 2017). Accordingly, our results show that prevention of 5′-cap trimethylation by TGS1 inhibition precludes exosome-mediated decay of both long extended hTR precursors and mature hTR molecules in both wild-type and DKC1 mutant cells. Surprisingly, the increased stability of hTR upon inhibition of 5′-cap trimethylation happens despite a concomitant significant increase in 3′-end oligoadenylation of hTR molecules, which traditionally acts as a potent degradation signal for this noncoding RNA by the exosome. We show that TGS1 binds to long extended hTR precursors prior to PAPD5, indicating a stepwise sequence of events where 5′-end methylation status initially modulates decay of 3′-end long extended hTR precursors, and then 3′-end oligoadenylation regulates decay of short extended and mature hTR molecules, as long as these remain 5′-cap-trimethylated. The combined inhibition of 5′-end TMG cap formation and 3′-end oligoadenylation rescues hTR expression and telomerase activity in cells harboring different pathogenic mutations in telomerase back to WT levels, including in hematopoietic CD34⁺/CD45⁺ progenitors, highlighting the potential of regulation of these pathways to be explored for treatment of bone marrow failure in TBD patients.

Results

5′-End cap trimethylation and 3′-end oligoadenylation modulate hTR decay rates independently, synergistically impacting mature hTR levels

Most mutations found in TBD patients disrupt hTR processing, causing rapid decay and reduced levels of mature hTR molecules (Armanios and Blackburn 2012; Savage 2014). This accelerated decay impairs the assembly of the telomerase complex and results in severe clinical manifestations (Armanios and Blackburn 2012; Savage 2014; Alter et al. 2015). This large number of TBD patients can therefore be classified as suffering from a ribonucleoprotein assembly disease, where the decay of hTR outpaces the assembly of telomerase. Increased expression of hTR rescues this phenotype (Agarwal et al. 2010; Batista et al. 2011), and we and others have shown that the reduction of hTR 3′-end oligoadenylation by PAPD5 inhibition or the prevention of hTR 5′-cap trimethylation by TGS1 silencing increases levels of mature hTR in telomerase mutant cells (Boyraz et al. 2016; Fok et al. 2019; Chen et al. 2020; Nagpal et al. 2020). However, this rescue remains akin to hTR levels found in wild-type (WT) cells (Shukla et al. 2016, 2020).

To understand whether the protective effect of PAPD5 and TGS1 inhibition on hTR happens independently or through modulation of a converging decay pathway, we initially silenced both genes either individually or in combination in WT and dyskerin mutant (DKC1_ΔL37) human fibroblasts. hTERT was exogenously expressed in these fibroblasts to allow for continuous growth and correct telomerase assembly (Supplemental Fig. S1A [hTERT levels], B [silencing efficiency]). ΔL37 is a common and severe mutation in patients with telomere biology disorders, showing only 15%–20% of hTR levels when compared with WT cells and significantly reduced telomerase activity (Batista et al. 2011). While the independent or combined silencing of TGS1 and PAPD5 in WT cells caused a small increase in hTR levels (Supplemental Fig. S1C, left panel, 9 days of treatment), its effect is significantly exacerbated in dyskerin mutant DKC1_ΔL37 cells (Fig. 1A, qPCR; Supplemental Fig. S1D, Northern blot), where hTR is otherwise quickly degraded (Batista et al. 2011). These data also demonstrate that the combined silencing of TGS1 and PAPD5 causes a significantly higher upregulation of hTR compared with the silencing of each of these genes independently (Fig. 1A; Supplemental Fig. S1D, Northern blot). To confirm these results, we chemically inhibited TGS1 and PAPD5 using sinefungin (Galati et al. 2022) and RG7834 (Mueller et al. 2018), respectively. WT and DKC1_ΔL37 mutants were treated with DMSO, 30 µM sinefungin, 5 µM RG7834, or both compounds for 20 days, at which point hTR levels were then assessed (please note that this treatment regimen was followed throughout this study). Chemical inhibition of TGS1 and PAPD5 recapitulated the genetic silencing results obtained in WT cells, with moderate changes to hTR levels (Supplemental Fig. S1C, right panel). Again, in sharp contrast to WT, treatment of DKC1_ΔL37 cells with either sinefungin or RG7834 caused a significant increase in levels of hTR (Fig. 1B, qPCR; Supplemental Fig. S1D, Northern blot). Additionally, the combined treatment of these cells with sinefungin and RG7834 had a synergistic effect in hTR levels, showing a significantly higher increase when compared with treatment with each drug individually (Fig. 1B; Supplemental Fig. S1D). As sinefungin is a broad-spectrum methyltransferase inhibitor, we next decided to establish whether its action on hTR levels is happening through inhibition of TGS1 activity. Treatment with sinefungin and siTGS1 independently leads to similar increases in hTR levels (Supplemental Fig. S2A), and the combined treatment does not produce any additive or synergistic effects, inferring that sinefungin's impact on hTR occurs primarily through inhibition of TGS1 rather than through off-target methyltransferases. To substantiate this conclusion, we generated a DOX-inducible CRISPR interference (CRISPRi) HeLa cell line harboring two sgRNAs targeting the endogenous TGS1 transcription start site (sgTGS1) (see the Materials and Methods for details). Upon DOX induction at 0.5 µg/mL for 9 days, sgTGS1 cells displayed TGS1 repression (Supplemental Fig. S2B), which was accompanied by a significant increase in hTR (Fig. 1C). To confirm that this effect requires the enzymatic activity of TGS1, we performed a rescue experiment by transiently expressing either WT or catalytically impaired TGS1 mutants D696A and W766A (Mouaikel et al. 2003; Monecke et al. 2009). After 9 days under DOX treatment, WT TGS1 was able to restore hTR levels to baseline, whereas expression of either mutant was not (Fig. 1C), indicating that hTR levels are regulated by the catalytic activity of TGS1. These results reveal that inhibition of TGS1 activity, when combined with inhibition of PAPD5, leads to a synergistic effect in the regulation of hTR levels, indicating that 5′-cap trimethylation and 3′-end oligoadenylation act independently during the process of hTR maturation.

Figure 1.

TGS1 and PAPD5 synergistically regulate hTR levels and telomerase activity in telomerase mutant cells. (A) Quantification of hTR levels by quantitative reverse transcription polymerase chain reaction (RT-qPCR) in wild-type (WT) and DKC1_ΔL37 cells after 9 days of siRNA-mediated knockdown of TGS1, PAPD5, or both (combo). (B) Quantification of hTR levels by RT-qPCR in WT and DKC1_ΔL37 cells after 20 days of treatment with sinefungin (TGS1 inhibitor), RG7834 (PAPD5 inhibitor), or both (combo). For A and B, data are presented as mean ± SEM (n ≥ 3), normalized to GAPDH, and shown as fold changes relative to wild-type conditions (WT = 1). Statistical significance was assessed by one-way ANOVA with Tukey's multiple comparisons test. (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001. (C) RT-qPCR quantification of hTR levels in CRISPRi TGS1 HeLa cells transfected with either wild-type TGS1 (WT) or catalytic mutants (D696A and W766A). Data represent mean ± SEM from n = 4 biological replicates. Statistical significance was determined by one-way ANOVA followed by Šídák's multiple comparisons test. (****) P < 0.0001, (ns) not significant. (D) hTR decay kinetics following actinomycin-D treatment in WT and DKC1_ΔL37 cells under the indicated conditions. hTR levels were analyzed at the indicated time points after treatment. Data are normalized to 18S rRNA and presented as mean ± SEM (n = 3). Statistical significance is shown in Supplemental Figure S2C. (E) Telomerase activity by telomere repeat amplification in WT and DKC1_ΔL37 cells after genetic silencing (left) or chemical inhibition (right) of TGS1 and PAPD5. The range of protein concentrations for each condition represents twofold serial dilutions (2, 1, and 0.5 µg). (LC) Loading control. (F) Schematic representation of primer positions used to distinguish different hTR species. Amplicon A detects mostly mature (as it can also detect residual extended precursor) hTR molecules (93–166 nt), and amplicons B (355–610 nt), C (452–610 nt), and D (722–896 nt) detect exclusively hTR precursors, which are progressively longer. Primer sequences are detailed in Supplemental Table S1. (G) RT-qPCR analysis of different hTR molecules (amplicons A–D) from TGS1 immunoprecipitated (IP) RNA in DKC1_ΔL37 cells. Data are expressed as percentage input and shown as fold enrichment relative to IgG controls. GAPDH served as a negative control, and U3 snoRNA served as a positive control. Mean ± SEM; n ≥ 3. Statistical analysis was performed using Mann–Whitney test. (*) P < 0.05; (ns) not significant. (H) TMG cap immunoprecipitation (IP) of mature and extended hTR molecules from DKC1_ΔL37 cells using anti-TMG antibodies (IgG as control). RT-qPCR was performed for amplicons A–D from immunoprecipitated RNA. Data are expressed as percentage input and shown as fold enrichment relative to IgG controls. GAPDH served as a negative control. Mean ± SEM; n ≥ 3. Statistical analysis was performed using Mann–Whitney test. (*) P < 0.05; (ns) not significant. (I) Enrichment of mature hTR (amplicon A) and 3′-extended precursor hTR (amplicons B–D) species following PAPD5 immunoprecipitation (IP). Twin-strep:3xFLAG:PAPD5 or Twin-strep:3xFLAG:mCherry (control) was transiently expressed in HEK293T cells. Twin-strep:3xFLAG:PAPD5 or Twin-strep:3xFLAG:mCherry was purified using FLAG beads followed by Strep beads, and associated RNA was analyzed by RT-qPCR. Data were normalized to U1 and represent fold enrichment of amplicons A–D in PAPD5 pull-downs compared with mCherry controls, presented as mean ± SEM (n ≥ 3). Statistical analysis was performed using two-tailed t-test. (**) P < 0.01; (ns) not significant.

To elucidate whether the observed increase in hTR levels upon combined inhibition of TGS1 and PAPD5 in dyskerin mutant cells is caused by augmented hTR stability rather than increased transcription rates, we quantified decay kinetics of hTR in different conditions. Experiments where RNA polymerase II (and therefore hTR transcription) is inhibited with actinomycin-D for different amounts of time confirmed that hTR is quickly decayed in DKC1_ΔL37 mutants compared with WT (6.2 vs. 14.6 h, respectively) (Fig. 1D; statistical significance shown in Supplemental Fig. S2C). Individual inhibition of either PAPD5 (cells treated with RG7834) and TGS1 (cells treated with sinefungin) in DKC1_ΔL37 mutants resulted in a noticeable increase in hTR stability, increasing its half-life to 8.9 and 7.7 h, respectively (Fig. 1D). The combined inhibition of PAPD5 and TGS1 brought the half-life of hTR in DKC1_ΔL37 mutants close to the half-life observed in WT cells, to 11.3 h (Fig. 1D). These results show that increased hTR levels observed in mutant cells after combined inhibition of PAPD5 and TGS1 (Fig. 1A,B) are likely caused by increased stability of hTR transcripts rather than increased transcription rates. Next, to verify whether the resulting hTR molecules are functional after TGS1 and PAPD5 inhibition, we assessed telomerase activity by telomerase repeat amplification protocol (TRAP). TRAP analysis shows that both the genetic silencing (Fig. 1E, left panel) and chemical inhibition (Fig. 1E, right panel) of PAPD5 and TGS1, both independently and in combination, increase telomerase activity in DKC1_ΔL37 mutants. The combined inhibition of TGS1 and PAPD5 led to a significantly higher rescue of telomerase activity in DKC1_ΔL37 mutants, to levels approximating those observed in WT cells (see TRAP quantitation in Supplemental Fig. S2D). Cotreatment of siTGS1 and sinefungin did not produce any additive increase in TRAP activity (Supplemental Fig. S2E), confirming that sinefungin acts primarily through TGS1 inhibition. Finally, as reduced hTR levels in dyskerin mutant cells cause severe hematopoietic failure (Fok et al. 2017), we analyzed whether the combined inhibition of TGS1 and PAPD5 can restore levels of hTR in hematopoietic lineages. Utilizing an in vitro platform of hematopoietic development from human embryonic stem cells (see the Materials and Methods) that recapitulates the major steps of in vivo blood development (Fok et al. 2019; Jeong et al. 2023), we measured hTR expression in hematopoietic progenitors (CD34⁺/CD45⁺) carrying a DKC1_A353V mutation (a common mutation found in TBD patients presenting with bone marrow failure). CD34⁺/CD45⁺ progenitor cells were chosen for analysis because this population is telomerase-positive and is severely compromised in TBD patients (Fok et al. 2017). Combined inhibition of TGS1 and PAPD5 significantly increased hTR levels in DKC1 mutant CD34⁺/CD45⁺ hematopoietic progenitors, largely surpassing levels found in isogenic WT counterparts (Supplemental Fig. S3A). Together, these results indicate that the combined inhibition of 5′ TMG capping and 3′-end oligoadenylation has a synergistic effect in preventing decay of hTR across different cellular populations.

TGS1 acts upstream of PAPD5 to mark extended hTR precursors for processing

Following transcription by RNA polymerase II, hTR molecules undergo 3′-end trimming by 3′–5′ exonucleases to reach their mature 451 nt form (Tseng et al. 2015). Nascent hTR transcripts can extend up to 1000 nt, and while the mechanisms regulating their processing remain incompletely understood (Tseng et al. 2015; Roake et al. 2019), current evidence suggests that the balance between exosome-mediated degradation and processing of precursor hTR molecules is a major factor determining the steady-state levels of mature hTR molecules (Moon et al. 2015; Nguyen et al. 2015; Tseng et al. 2015). In this context, we wanted to understand how the inhibition of TGS1, which prevents the trimethylation of the 5′ cap of hTR, influences this process, as it prevents hTR decay by itself and amplifies the effect of PAPD5 inhibition (Fig. 1A,B).

To address this question, we initially determined the pattern of binding of TGS1 and PAPD5 to early transcribed (long extended) hTR precursors or mature/short extended hTR molecules. We performed immunoprecipitation of endogenous TGS1 from both WT and DKC1_ΔL37 cells (immunoprecipitation efficiency shown in Supplemental Fig. S3B) followed by amplification of hTR molecules using sets of oligos that recognize different regions of hTR (model in Fig. 1F). This strategy allows the determination of different species of hTR molecules that are bound by TGS1. The regions analyzed include the mature hTR sequence (amplicon A [93–165 nt], which can also detect residual 3′-end extended precursors that have not been processed/decayed) and different sets of oligos that exclusively target progressively longer 3′-end genomically extended regions of hTR, including amplicon B (355–610 nt), amplicon C (451–610 nt), and amplicon D (772–896 nt). Enrichment analysis by RT-qPCR performed from the immunoprecipitated fraction shows that TGS1 binds to all hTR species analyzed in both DKC1_ΔL37 (Fig. 1G) and WT cells (Supplemental Fig. S3C). To understand whether TGS1 is catalytically active on these long extended hTR precursors, we specifically isolated RNAs with trimethylated (TMG) 5′ caps from both WT and DKC1_ΔL37 cells utilizing an antibody that specifically recognizes TMG caps. In both WT (Supplemental Fig. S3D) and DKC1_ΔL37 (Fig. 1H) cells, amplicon A, representing mostly mature hTR (and possibly extended sequences still present), was enriched in TMG immunoprecipitated samples relative to the IgG control. As a control, GAPDH, which carries a monomethylguanosine (MMG) cap at its 5′ end (Jurado et al. 2014), did not show enrichment in the TMG immunoprecipitated samples (Fig. 1H; Supplemental Fig. S3D). Notably, amplicons B–D, which capture exclusively 3′-extended hTR precursors, also show enrichment in TMG immunoprecipitated samples in both WT (Supplemental Fig. S3D) and DKC1_ΔL37 (Fig. 1H) cells. These amplicons (in both WT and mutant cells) were independently analyzed by gel electrophoresis of semiquantitative PCR products (Supplemental Fig. S3E), which confirmed that long genomically extended hTR molecules undergo 5′-cap trimethylation prior to processing or decay. Next, we wanted to determine the pattern of interaction of PAPD5 with these different hTR molecules (long extended precursors and mature/short extended molecules). For technical reasons, we could not perform the PAPD5 IP experiments in primary fibroblasts (see the Materials and Methods). Therefore, we transiently transfected HEK293T cells with plasmids encoding either Twin-strep:3xFLAG:PAPD5 or Twin-strep:3xFLAG:mCherry as a negative control. Following transfection, we performed tandem affinity purification, initially using FLAG beads and subsequently using Strep beads (immunoprecipitation efficiency shown in Supplemental Fig. S4A). Unlike what we observed for TGS1, RT-qPCR analysis of hTR molecules from PAPD5 immunoprecipitated samples shows no enrichment of 3′-end extended precursors (amplicons B–D), only of mature hTR molecules (amplicon A) (Fig. 1I). These results indicate a distinct action of TGS1 and PAPD5 during hTR maturation, which we initially sought to determine by analyzing the pattern of binding of TGS1 and PAPD5 to specific species of hTR molecules. We performed hTR 3′-rapid amplification of cDNA ends (RACE) analysis on RNA extracted from both TGS1 and PAPD5 immunoprecipitates (Supplemental Fig. S4B, model for the 3′-RACE strategy). We saw a stronger interaction of TGS1 with long extended hTR (>168 bp) compared with mature hTR (168 bp) molecules (Supplemental Fig. S4C, left image). In contrast, PAPD5 associates with mature but not with long genomically extended forms of hTR (Fig. 4C, right image). As an additional control, we confirmed recent nascent RNA sequencing data that indicate that short extended 3′-end precursors are oligoadenylated (Roake et al. 2019) and we show that PAPD5 immunoprecipitated samples are enriched for short hTR precursors that are up to at least 21 nt past position 451 (Supplemental Fig. S4D, RPLP0 was used as a negative control). Together, these results demonstrate that while PAPD5 mostly interacts with mature hTR and short 3′-extended precursors, TGS1 binds to long genomically extended hTR precursors early in the biogenesis pathway. This suggests a sequential mode of action for the processing of hTR, with TGS1 trimethylating the 5′ cap of long 3′-end genomically extended hTR molecules prior to 3′-end oligoadenylation of short extended and mature hTR molecules by PAPD5.

Figure 4.

Exosome licensing by 5′-cap trimethylation happens independently from PAPD5. (A) RT-PCR analysis of hTR amplicons A–D in iTGS1 cells cultured with or without 1 µg/mL doxycycline (DOX) for 20 days for induction of TGS1 and simultaneously treated for 20 days with DMSO (control) or 5 µM PAPD5 inhibitor RG7834. Bars represent mean ± SEM from n ≥ 4 independent experiments. Statistical analysis was performed using one-way ANOVA followed by Šídák's multiple comparisons test. (B) RT-qPCR quantification of 3′-extended hTR precursor species (amplicons B–D) in DKC1_ΔL37 cells in the indicated conditions. EXOSC3 silencing (siEXOSC3) was performed for the final 3 days prior to RNA collection. Data are presented as fold change in C_t values (ΔC_t) normalized to GAPDH (mean ± SEM, n ≥ 3). Statistical analysis was performed using two-way ANOVA followed by Šídák's multiple comparisons test. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001; (ns) not significant.

To validate this model, we examined whether PAPD5 inhibition affects TGS1-mediated trimethylation of the 5′-cap structure. We immunoprecipitated TMG-capped RNA from total RNA isolated from WT and DKC1_ΔL37 mutant fibroblasts, where TGS1 and PAPD5 were inhibited either independently or in combination, followed by RT-qPCR analysis. Despite the substantial increase in total hTR levels upon PAPD5 inhibition (Fig. 1A,B), the fraction of hTR molecules carrying a trimethylated 5′ cap remained unchanged (Supplemental Fig. S4E). As expected, the TMG-capped fraction of hTR molecules is significantly reduced in samples where TGS1 is inhibited either alone or in combination with PAPD5 (Supplemental Fig. S4E). Combined, the data presented here show that hTR precursors comprise a large amount of the molecules to which TGS1 is bound and that this binding likely precedes 3′-end oligoadenylation by PAPD5, which only interacts with mature and short extended hTR molecules.

Inhibition of 5′ TMG cap formation leads to accumulation of 3′-end oligoadenylated hTR molecules

The processing of hTR precursors into 451 nt mature hTR molecules happens slowly, which facilitates precursor degradation by RNA surveillance mechanisms (Tseng et al. 2015; Roake et al. 2019). This balance between degradation and maturation makes it challenging to accurately discern the length and processing rate of hTR immediately following transcription until loading onto telomerase. This difficulty is exacerbated in dyskerin mutants that show fast decay of this noncoding RNA. To explore whether TGS1 activity impacts the stability and processing of hTR at different stages (extended precursors and mature 451 nt molecules), we employed RNA ligase-mediated rapid amplification of cDNA ends in combination with high-throughput sequencing of hTR (3′-end RACE-seq). This method allows for precise mapping of hTR 3′ ends, enabling the identification of both variations in length and posttranscriptional modifications of hTR molecules at high resolution (Moon et al. 2015; Nguyen et al. 2015; Tseng et al. 2015; Roake et al. 2019). By utilizing 3′-end RACE-seq, we aimed to determine whether inhibition of TGS1 (and therefore prevention of 5′-cap trimethylation) affects the processing of hTR at its 3′ terminus, allowing for accumulation of longer forms of hTR precursors and thereby changing the proportion of mature versus extended transcripts.

3′-End RACE-seq analyses showed that in WT cells, ∼88.37% ± 0.47% of hTR molecules are mature (aligning at position 451), while 11.80% ± 0.24% extend beyond this point (Fig. 2A). Of the total reads, 1.38% ± 0.05% represent genomic extensions (>452 nt) at the 3′ end. As an adenosine is naturally encoded at position 452, it is not possible to differentiate between genomic extension and monoadenylation at this position, with 3.23% ± 0.31% of hTR molecules falling into this ambiguous category. Additionally, 7.02% ± 0.12% of total hTR molecules in WT cells are 3′-end oligo(A)-tailed, with 4.13% ± 0.03% of those tails found in mature hTR molecules and 2.89% ± 0.08% of those tails found on genomically extended 3′-end precursors (Fig. 2A, quantification in F). In contrast, DKC1_ΔL37 mutant cells show an increase in 3′-end oligo(A)-tailed hTR molecules (10.95% ± 1.41%) (Fig. 2B, quantification in F), with this increase affecting both mature (6.25% ± 0.51%) and genomically extended (4.70% ± 0.9%) forms. This increase likely stems from the absence of functional DKC1, making the 3′ end of hTR more susceptible to PAPD5-mediated adenylation and subsequent decay. Analysis of 3′-end RACE data in DKC1_ΔL37s also shows an increase in genomically extended hTR molecules (2.74% ± 0.77%), indicating a slower maturation rate in these mutants (Fig. 2B). As expected, inhibition of PAPD5 in DKC1_ΔL37 cells leads to a substantial reduction in the fraction of oligo(A)-tailed hTR species (to 2.19% ± 0.3%), which affects both mature and 3′-extended forms of hTR (Fig. 2C, quantification in F). Surprisingly, inhibition of TGS1, despite increasing hTR levels (Fig. 1A,B), causes a significant increase in the fraction of oligoadenylated hTR species, from 10.95% ± 1.41% to 19.20% ± 0.23% of total molecules (Fig. 2D, quantification in F). Most of this increase was observed in mature (451 nt) hTR, accounting for 14.09% ± 0.37% of total molecules, with only 5.11% ± 0.13% of adenylation occurring in genomically extended hTR forms (Fig. 2F). These results reveal that while the inhibition of TGS1 is not directly involved in the 3′-end maturation of hTR, it is able to prevent decay of 3′-end oligo(A)-tailed hTR molecules, which are normally targeted for degradation. This hypothesis is consistent with the increased hTR levels and telomerase observed after inhibition of TGS1 activity (Fig. 1). The length distribution of 3′-end oligo(A) tails specifically at position 451 (Fig. 2G) shows increased oligo(A) tail length in TGS1-inhibited cells. The dual inhibition of TGS1 and PAPD5 (Fig. 2E) shows a similar 3′-end distribution of hTR molecules when compared with inhibition of PAPD5 alone, with a similar reduction in the fraction of oligo(A)-tailed species (2.62% ± 0.09%). However, the dual inhibition of TGS1 and PAPD5 rescues hTR to levels significantly higher than the inhibition of PAPD5 alone (Fig. 1). This suggests that while 3′-end oligoadenylation leads to hTR degradation, this process is dependent on the methylation status of the 5′ cap of hTR. When both triggers of decay were disrupted (5′-cap trimethylation and 3′-end oligoadenylation), a pronounced, additive increase in functional hTR levels was observed.

Figure 2.

Inhibition of 5′ TMG cap formation leads to accumulation of 3′-end oligoadenylated hTR molecules. (A–E) Pie charts showing the distribution of hTR 3′-end positions, as determined by RNA ligase-mediated 3′-end RACE followed by high-throughput sequencing in the following samples: wild-type cells (A), DKC1_ΔL37 cells (B), and DKC1_ΔL37 cells treated with the PAPD5 inhibitor RG7834 (C), the TGS1 inhibitor sinefungin (D), or both (E). The mature 451 nt form is shown in gray. Genomically extended hTR species without nontemplated A tails are shown in light orange, with position 452 highlighted in orange. Nontemplated A-tailed forms are categorized as mature hTR with A tails (yellow) and >452 nt extended hTR with A tails (blue). The total adenylated fraction is represented in green. (F) Relative abundance of adenylated 3′-end reads in hTR under the indicated conditions. The number of adenylated reads was normalized to total hTR reads for each condition. Data are presented as mean ± SEM. Statistical significance was determined by one-way ANOVA with Tukey's multiple comparisons test. (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001, (ns) not significant. (G) Length distribution of oligo(A) tails at the mature 451 nt position of hTR under the indicated conditions. The fraction of reads containing two or more nontemplated adenosine residues was calculated relative to the total hTR reads for each condition. Data are presented as mean ± SEM.

MTR4 requires trimethylated 5′ caps to engage hTR precursors for exosome-mediated decay

Data presented so far show that long, genomically extended hTR precursors physically interact with TGS1 (Fig. 1G; Supplemental Fig. S3C) and that the inhibition of TGS1 stabilizes hTR levels (Fig. 1) despite an increase in 3′-end oligoadenylation at the 451 nt position (Fig. 2), which traditionally acts as a degradation signal. These results prompted us to investigate whether preventing 5′-cap trimethylation affects the abundance and stability of long 3′-extended hTR precursors regardless of their 3′-end oligoadenylation status, which could allow for their processing into mature molecules.

While amplicon A (composed mostly of mature hTR molecules) is significantly reduced in DKC1_ΔL37 cells when compared with WT cells (Fig. 1A,B), this reduction was not observed in long extended hTR precursors (Figs. 1F [model], 3A [amplicons B–D]). Surprisingly, chemical inhibition of TGS1 resulted in a significant accumulation of all extended hTR forms analyzed in both WT (Supplemental Fig. S5A) and DKC1_ΔL37 (Fig. 3A) cells. This accumulation occurred when TGS1 was inhibited both independently and in combination with PAPD5. In contrast, PAPD5 inhibition alone had no impact on extended hTR precursor levels in either WT (Supplemental Fig. S5A) or DKC1 mutant (Fig. 3A) cells. Genetic knockdowns of TGS1 and PAPD5 confirmed these results (Supplemental Fig. S5B), suggesting that TGS1 inhibition stabilizes long 3′-end extended hTR precursors, which are normally targeted for degradation. To further support this observation, we established a doxycycline-inducible TGS1 (iTGS1) HeLa cell line (Twin-strep:3xFLAG:TGS1). Induction of TGS1 upon DOX treatment shows a dose-dependent response in TGS1 expression levels (Supplemental Fig. S5C). Concomitantly, hTR levels across all hTR species analyzed were progressively reduced with increasing DOX concentrations (0, 0.25, 0.5, and 1.0 µg/mL) over 20 days (Fig. 3B; statistical significance shown in Supplemental Fig. S5D).

Figure 3.

MTR4 requires trimethylated 5′ caps to engage hTR precursors. (A) Quantification of 3′-extended hTR precursor molecules under the indicated conditions. RT-qPCR analysis was performed for amplicons B–D (see model in Fig. 1F) in wild-type (WT) and DKC1_ΔL37 cells treated for 20 days with DMSO (control), 30 µM sinefungin (TGS1 inhibitor), 5 µM RG7834 (PAPD5 inhibitor), or both. Data are presented as fold change in C_t values (ΔC_t) normalized to GAPDH (mean ± SEM; n ≥ 5). Statistical analysis was performed using one-way ANOVA with Tukey's multiple comparisons test. (*) P < 0.05, (**) P < 0.01, (****) P < 0.0001, (ns) not significant. (B) RT-qPCR analysis of hTR levels across amplicons A–D in iTGS1 cells treated with increasing concentrations of DOX (0, 0.25, 0.5, and 1.0 mg/mL). Statistical significance was assessed by one-way ANOVA with Dunnett's multiple comparisons test as shown in Supplemental Figure S5D. (C) Decay kinetics of 3′-extended hTR precursor molecules following actinomycin-D treatment in DKC1_ΔL37 cells. Expression levels of amplicons B–D were analyzed at the indicated time points following actinomycin-D treatment in cells treated with either DMSO (control) or sinefungin. Data are normalized to 18S rRNA and presented as mean ± SEM (n = 3). Statistical significance was assessed using two-way ANOVA followed by uncorrected Fisher's LSD test as shown in Supplemental Figure S5E. (D,E) Enrichment of mature hTR (amplicon A) and 3′-extended precursor hTR species (amplicons B–D) following MTR4 immunoprecipitation (IP). (D,E) Twin-strep:3xFLAG:MTR4 or Twin-strep:3xFLAG:mCherry (control) were transiently expressed in HEK293T cells after DKC1 knockdown (siDKC1) either individually or combined with PAPD5 silencing (D) or TGS1 silencing (E). MTR4 (or control mCherry) was purified using Strep beads, and associated RNA was analyzed by RT-qPCR. Data were normalized to GAPDH and represent fold enrichment of amplicons A–D in MTR4 pull-downs from double-knockdown samples (blue) relative to siDKC1 alone (black) and are shown as mean ± SEM (n = 3). Statistical analysis was performed using two-tailed t-test. (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001, (ns) not significant.

To directly test whether TGS1 modulates the stability of extended hTR precursors, we measured decay kinetics using actinomycin-D transcriptional inhibition. Analysis of decay kinetics of different extended hTR precursors shows that the inhibition of TGS1 activity significantly prolonged their half-lives (Fig. 3C; statistical significance shown in Supplemental Fig. S5E). These results indicate that TGS1 inhibition enhances the stability of long extended hTR precursors even in cells harboring a pathogenic mutation in DKC1. Together, our data suggest that the loss of 5′-cap trimethylation impairs the RNA decay machinery, preventing degradation of long extended hTR species. Given that hTR molecules are typically degraded by the exosome complex (Chen and Batista 2025), we hypothesized that 5′-cap trimethylation could influence exosome recruitment to hTR precursors.

To test this, we first examined the interaction between MTR4 (a key helicase of the NEXT complex responsible for directing RNAs to the exosome) (Lingaraju et al. 2019; Schmid and Jensen 2019; Gerlach et al. 2022) and different hTR species. We transiently expressed Twin-strep:3xFLAG:MTR4 or Twin-strep:3xFLAG:mCherry (control) in HEK293T cells (the number of cells needed for these assays prevented the use of fibroblasts) followed by immunoprecipitation of the tagged proteins. RNA extracted from pull-down samples was then analyzed for the presence of different hTR species. The specificity of the MTR4 pull-down was confirmed by the coimmunoprecipitation of EXOSC3, a core exosome component, which was absent in mCherry controls (Supplemental Fig. S6A). Our analysis shows that MTR4 efficiently interacts with hTR at multiple stages of processing, as pull-downs contained both mature hTR (detected by amplicon A) and long 3′-extended hTR precursors (detected by amplicons B–D) (Supplemental Fig. S6B). We next evaluated how the inhibition of TGS1 or PAPD5 influences this interaction. We performed MTR4 pull-down assays in cells where DKC1 was silenced (to allow for increased hTR decay) followed by depletion of either PAPD5 or TGS1 (knockdown efficiencies are shown in Supplemental Fig. S6C). While the inhibition of PAPD5 reduced the interaction of MTR4 exclusively with mature hTR molecules and did not influence its binding to hTR precursors (Fig. 3D), silencing of TGS1 significantly impaired the binding of MTR4 to both mature and 3′-end extended hTR precursors (Fig. 3E). These results establish a critical role of 5′-cap trimethylation in facilitating exosome recruitment via MTR4 to hTR precursors, whereas PAPD5 primarily modulates the interaction of MTR4 with mature (and possibly short extended) hTR molecules.

To further establish this parallel action of PAPD5 and TGS1 during hTR processing, we tested whether inhibition of PAPD5 can restore levels of different hTR species in DOX-positive iTGS1 cells. Analysis of Figure 4A shows that PAPD5 inhibition is unable to rescue hTR levels when TGS1 is overexpressed. These results strongly support a scenario where the trimethylation of 5′ caps by TGS1 is an essential step for the decay of both precursor and mature hTR molecules by the exosome independently and prior to their 3′-end oligoadenylation by PAPD5. To corroborate this hypothesis and directly test whether 5′-cap trimethylation is essential for exosome-mediated degradation, we silenced EXOSC3 (silencing efficiency shown in Supplemental Fig. S6D) in cells where TGS1 and PAPD5 were inhibited either independently or in combination. To minimize possible secondary effects (Gockert et al. 2022), EXOSC3 silencing was performed only for the final 3 days of the experiment. If 5′-cap trimethylation is indeed an essential element driving the exosome-mediated decay of extended hTR precursors, the depletion of EXOSC3 should cause the accumulation of these precursors in both control and PAPD5-inhibited cells but not when TGS1 is inhibited and the 5′ caps remain monomethylated. Consistent with this hypothesis, EXOSC3 depletion leads to the accumulation of long 3′-extended precursors in control and PAPD5-inhibited cells but not in cells where TGS1 is inhibited either by itself or in combination with PAPD5 (Fig. 4B). These data support the hypothesis that 5′-cap trimethylation by TGS1 determines the recruitment of the exosome complex to long extended hTR molecules, promoting their decay.

Finally, Northern blot analysis of hTR levels (Supplemental Fig. S6E) confirmed that PAPD5 and TGS1 have distinct roles during hTR maturation, as there was a significant increase in mature hTR levels when either was inhibited, with an even greater visible increase when both were concomitantly inhibited (Supplemental Fig. S6E, lane 2 vs. lanes 3,4,5). Unlike what was observed in precursor hTR molecules (Fig. 4B), inhibition of TGS1 in cells with silenced EXOSC3 resulted in increased levels of mature hTR (Supplemental Fig. S6E, lane 3 vs, lane 7). If, on top of TGS1 inhibition, we also inhibited PAPD5 in EXOSC3 silenced samples, there was no significant increase in mature hTR levels compared with the simple silencing of EXOSC3 (Supplemental Fig. S6E, cf. lanes 5 and 9), as combined inhibition of these proteins completely prevented exosome targeting of hTR (Supplemental Fig. S6E).

Combined, these results show a stepwise processing and turnover of hTR molecules, where their decay is modulated posttranscriptionally and epitranscriptionally at different stages of maturation. While TGS1-mediated 5′-cap trimethylation is essential for exosome recruitment (and decay) of long hTR precursors independently of their 3′-end adenylation, both modifications (namely, 5′-cap trimethylation and 3′-end oligoadenylation) regulate the decay of mature hTR molecules by the exosome.

Altered cellular compartmentalization prevents exosome-mediated decay of 5′ MMG-capped hTR precursors

Depletion of TGS1 causes accumulation of mature hTR molecules in the cytoplasm (Chen et al. 2020). This suggests a potential mechanism for hTR precursors to escape exosome-mediated degradation through migration from nuclear exonucleases. To examine how TGS1 inhibition affects the localization of its primary targets (extended hTR precursors), we initially performed nuclear–cytoplasmic fractionation to assess extended hTR precursor distribution after TGS1 or PAPD5 inhibition (individually or in combination). The efficiency of nuclear–cytoplasmic fractionation was validated through expression of different GAPDH transcripts (Supplemental Fig. S7A), including GAPDH precursors and introns (nuclear markers) and GAPDH exons and actin exons (cytoplasmic markers). Our fractionation experiments show a specific increase of hTR levels in the cytoplasmic fraction after TGS1 inhibition, which is restricted to amplicon A, mostly composed of mature hTR species. This result was also observed when TGS1 was inhibited in combination with PAPD5 (Fig. 5A, amplicon A). In contrast, PAPD5 inhibition does not elevate cytoplasmic hTR levels, indicating that 3′-end oligoadenylation does not influence trafficking of hTR molecules in the cell (Fig. 5A, amplicon A). Instead, PAPD5 inhibition led to a significant increase in mature hTR levels within the nucleus (Fig. 5B, amplicon A). Notably, despite the overall increase in hTR levels upon combined inhibition of TGS1 and PAPD5 (Fig. 1A,B), these fractionation analyses did not show a synergistic enhancement in cytoplasmic mature hTR levels upon dual inhibition of PAPD5 and TGS1 beyond the levels observed with TGS1 inhibition alone (Fig. 5A, amplicon A). This suggests that the synergistic stabilization of hTR observed upon dual inhibition of TGS1 and PAPD5 (Fig. 1A,B) primarily reflects nuclear retention rather than enhanced cytoplasmic export. These results argue that TGS1 inhibition promotes hTR stability by impairing exosome-mediated degradation in the nucleus and not by rerouting hTR to the cytoplasm. In sharp contrast to mature hTR (amplicon A), long 3′-end genomically extended hTR precursors show no enrichment in the cytoplasmic fraction under any condition, including the inhibition of TGS1 alone or in combination with PAPD5 (Fig. 5A, amplicons B–D). Rather, these fractionation analyses uncover that the overall increase observed in precursor hTR molecules after TGS1 inhibition happens through accumulation of these species in the nucleus (Fig. 5B, amplicons B–D). These results reveal that extended precursor forms of hTR are predominantly nuclear and do not undergo cytoplasmic transport regardless of whether or not their 5′-end cap is monomethylated (following TGS1 inhibition). The lack of cytoplasmic export for these precursors corroborates the notion that these molecules are primarily targeted for degradation by the nuclear exosome and that TGS1 inhibition, which avoids trimethylation of hTR's 5′ cap, prevents this process from happening.

Figure 5.

Altered cellular compartmentalization prevents exosome-mediated decay of 5′ MMG-capped hTR precursors. (A,B) Quantification of different hTR species in the cytoplasmic (A) and nuclear (B) fractions of DKC1_ΔL37 cells treated with DMSO (control), sinefungin (TGS1 inhibitor), RG7834 (PAPD5 inhibitor), or both (combo). Specific primer locations for amplicons A–D are detailed in Figure 1F. Data are shown as fold changes of ΔC_t values normalized to the GAPDH exon (cytoplasmic fraction) or GAPDH intron (nuclear fraction). Results represent the mean ± SEM from at least three independent experiments (n ≥ 3). For statistical analysis, amplicon A was analyzed using one-way ANOVA with Fisher's least significant difference (LSD) test, and amplicons B–D were analyzed using one-way ANOVA with Dunnett's multiple comparisons test. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001, (ns) not significant. (C) Representative RNA fluorescence in situ hybridization (FISH) images of wild-type (WT) and DKC1 silenced (siDKC1) cells treated with DMSO (control), sinefungin, RG7834, or both (combo). FISH was performed using Quasar 670-labeled probes that specifically recognize 3′-extended hTR molecules. Nuclei were stained with DAPI (blue), and nucleoli were visualized by immunostaining for fibrillarin (green). Scale bars, 10 µm. Insets represent zoomed-in images obtained from the highlighted areas. (D) Quantification of the average number of 3′-extended hTR foci per nucleus from RNA FISH experiments shown in C. (E) Violin plot representation of RNA FISH signal intensity (integrated fluorescence density per focus; arbitrary units [AUs]) of 3′-extended hTR foci from experiments shown in C. For D and E, 100 cells were scored per condition from three independent experiments. Statistical analysis was performed using one-way ANOVA with Dunnett's multiple comparisons test. (*) P < 0.05, (**) P < 0.01, (****) P < 0.0001, (ns) not significant.

Finally, we wanted to comprehensively define the subcellular localization of 3′-extended hTR precursors under different 5′-cap methylation statuses. For that, we performed fluorescence in situ hybridization (FISH) experiments using a set of 16 specific 20 nt long probes targeting exclusively the 3′-extended region of hTR (nucleotides 452–1316) (Supplemental Fig. S7B). FISH analysis in control samples (performed in HeLa cells due to technical reasons) confirmed previous reports (Nguyen et al. 2015) that extended hTR transcripts, at least transiently, localize to the nucleolus (Fig. 5C, fibrillarin used as a nucleolar marker). We then examined how 5′-cap methylation status and 3′-end oligoadenylation influence the localization and abundance of 3′-extended hTR species. FISH was performed across different conditions in both WT and DKC1-depleted cells (including cells treated with sinefungin, RG7834, or both). These analyses show significant nucleolar enrichment of 3′-extended hTR precursors upon TGS1 inhibition by itself or in combination with PAPD5 (Fig. 5C,D). The number of hTR foci increases significantly when TGS1 is inhibited (Fig. 5D). Also, quantification of fluorescence intensity reveals significantly higher integrated fluorescence density in TGS1-inhibited samples (alone or in combination with PAPD5 inhibition) compared with controls, reflecting increased abundance of extended hTR transcripts within these nucleolar foci (Fig. 5E). These results are consistent with both our expression analysis (Fig. 1) and fractionation experiments (Fig. 5A,B) and further support a model where TGS1-mediated cap trimethylation is a critical modification for recognition of extended hTR molecules by the exosome, as major components of the NEXT complex are excluded from the nucleolus. As the inhibition of PAPD5 alone does not cause a similar increase in the FISH signal for 3′-extended hTR species (Fig. 5C–E), these results further substantiate that the stabilization of extended hTR transcripts is not dependent on 3′-end oligoadenylation status despite PAPD5 and PARN being localized to the nucleolus (Fong et al. 2013; Ogami et al. 2013).

Discussion

Interplay between 5′-cap trimethylation and 3′-end oligoadenylation synergistically regulates hTR maturation and decay in DKC1 mutants

The biogenesis and assembly of telomerase is a complex and dynamic process that ensures the precise regulation of telomere length from early stages of development to aging (Revy et al. 2023; Chen and Batista 2025). Deficiencies in the balance between telomere elongation and telomere shortening are associated with human disease, highlighting the importance of this process. For example, patients harboring mutations in DKC1 have reduced levels of mature hTR and present with severe phenotypes. The dyskerin complex is essential for the accumulation of mature hTR (Fu and Collins 2007; Venteicher et al. 2008), as it inhibits the 3′-to-5′ degradation of hTR molecules near the mature, 451 nt, 3′ end (Tseng et al. 2015, 2018). hTR maturation is also influenced by the 3′-end deadenylation of short genomically extended hTR precursors by PARN, which promotes hTR stability (Moon et al. 2015; Tseng et al. 2015; Boyraz et al. 2016; Roake et al. 2019). Working in opposition to PARN, PAPD5 promotes the oligoadenylation of short genomically extended and mature molecules of hTR, causing its decay by the nuclear exosome complex. Here, we show that when combined with the inhibition of the trimethyl guanosine synthase TGS1, which trimethylates the 5′ cap of hTR, the inhibition of PAPD5 in DKC1 mutants increases hTR stability to levels comparable with those found in WT cells, including in human hematopoietic progenitor lineages. TGS1 immunoprecipitations show that TGS1 binds to 3′-end long extended hTR precursors; however, inhibition of TGS1 increases not only the stability of these long extended precursors but also levels of functional hTR. In contrast, inhibition of PAPD5 only rescues levels of mature/short extended hTR molecules, not affecting the stability or levels of long extended precursors. These results suggest that hTR biogenesis happens in a sequential but modular process. TGS1 activity targets both mature and long precursors to decay by the exosome, whereas PAPD5 acts on short/mature forms of hTR independently of TGS1, also targeting these for decay. PARN trims hTR short precursors, also preventing their degradation (Moon et al. 2015; Roake et al. 2019). Therefore, combined inhibition of TGS1 and PAPD5 creates a double layer of protection that is able to increase hTR levels and telomerase activity even in dyskerin mutant cells, which have extremely low levels of hTR. A model for the mechanisms regulating the processing and degradation of hTR is shown in Figure 6. It is tempting to speculate that modulating the activity of posttranscriptional regulators of 5′-end capping and 3′-end processing shifts the balance of hTR decay and processing. However, more experiments are needed to clarify the processes regulating decay versus maturation of hTR into functional molecules that allow for telomerase activity.

Figure 6.

Model describing TGS1 and PAPD5 regulation of hTR processing and exosome-mediated decay. hTR is initially transcribed as 3′-end long extended precursors. The fate of these transcripts is synergistically shaped by sequential 5′-cap trimethylation and 3′-end oligoadenylation. The top row shows hTR processing when both TGS1 and PAPD5 are present. In this scenario, long hTR precursors are 5′ TMG-capped by TGS1, leading to recruitment of MTR4 (a component of the nuclear NEXT complex) and the exosome, which induces hTR degradation and reduced processing into mature hTR molecules. The middle row shows hTR processing when TGS1 is absent but PAPD5 is present. Without TGS1, long genomically extended hTR precursors retain a 5′ MMG cap, which reduces MTR4/NEXT-mediated recruitment of the exosome, resulting in stabilization of these precursors. This increased stability may allow their processing (by PARN and/or other 3′–5′ exoribonucleases). PAPD5 oligoadenylation in the absence of 5′ trimethylation does not influence exosome targeting of hTR precursors despite increased 3′-end oligoadenylation. This leads to partial stabilization of mature hTR, increasing telomerase activity. Finally, the bottom row depicts hTR processing when both TGS1 and PAPD5 are absent. Here, monomethylated 3′-end long genomically extended precursors are not targeted by MTR4/NEXT and are protected from exosome action. Without PAPD5, these precursors can now be further processed to mature hTR molecules that are not targeted by the exosome. The combined absence of 5′ TMG capping and 3′ oligoadenylation synergistically increases mature hTR levels, greatly increasing telomerase RNA assembly. This model demonstrates that TGS1 promotes degradation of both long 3′ genomically extended precursors via MTR4/NEXT and mature hTR molecules, while PAPD5 targets exclusively 3′ short precursors and mature hTR species for degradation by the exosome via oligoadenylation. These processes control not only hTR biogenesis but also telomerase assembly and telomere maintenance in human cells, ultimately influencing cellular viability across the life span.

5′ MMG-capped hTR molecules are not recognized by the exosome despite increased 3′-end oligoadenylation

The role of long genomically extended molecules in the formation of mature, 451 nt hTR is unclear, as current data do not exclude the possibility that these molecules are able to escape decay and possibly be processed into mature forms of hTR (Tseng et al. 2015; Roake et al. 2019). Here, we show that inhibition of TGS1 stabilizes long extended precursors, leading to their accumulation. Prevention of 5′-cap trimethylation also increases levels of functional hTR molecules despite a concomitant increase in 3′-end oligoadenylation at the 451 nt position (Fig. 2). It is tempting to speculate that the observed increase in 3′-end adenylation at the 451 nt position following inhibition of TGS1 is derived from the lower MTR4 targeting of these molecules (Fig. 3E), as silencing of MTR4 has been shown to increase 3′-end adenylation at the 451 nt position (Tseng et al. 2015). The increased 3′-end oligoadenylation of hTR at the 451 nt position following prevention of 5′-cap trimethylation further demonstrates the dynamic process of hTR maturation, as it indicates that methylation levels of the 5′ cap influence processing of the opposing end of the telomerase RNA molecule. A similar phenotype has recently been suggested to occur during maturation of U2 snRNA, which implicates a broad role of TGS1 in the 3′-end processing of noncoding RNAs (Chen et al. 2022). Mechanistically, prevention of 5′ TMG cap formation reduces binding of the MTR4 helicase to both mature/short extended and long, genomically extended hTR molecules. These results indicate a direct role for the NEXT complex (composed of MTR4, ZCCH8, and RBM7) in the decay of long extended hTR precursors by the exosome. Supporting this, we show that when TGS1 is inhibited, silencing of the exosome component EXOSC3 does not further increase levels of genomically extended hTR molecules, establishing formation of a 5′ TMG cap as a necessary step for decay of hTR precursors. Furthermore, prevention of 3′-end adenylation by PAPD5 inhibition does not reduce the binding of MTR4 to long, genomically extended hTR precursors but to short extended/mature hTR molecules only. This indicates that 5′-cap trimethylation represents an early, initial step in the modulation of hTR decay followed by the regulation of 3′-end oligoadenylation. This hypothesis is further supported by analysis of cells harboring clinically relevant mutations in ZCCH8, which also show accumulation of genomically extended hTR (Gable et al. 2019).

Posttranscriptional regulation of hTR determines its subcellular localization

Cellular compartmentalization plays a central role during telomerase biogenesis, with proper hTR localization ultimately determining telomerase activity (Klump et al. 2023). While in normal settings hTR molecules are only transiently associated with the nucleolus, deletion of TCAB1 causes a significant increase of hTR in this cellular compartment, which impairs telomerase activity (Venteicher et al. 2009; Zhong et al. 2011; Klump et al. 2023). Interestingly, our cellular fractionation analysis shows that upon TGS1 inhibition, 3′-end extended hTR precursor levels are increased exclusively in the nucleus. Additionally, immunofluorescence analyses show increased levels of 3′-end extended hTR precursors in the nucleolus upon TGS1 inhibition, concomitant with increased telomerase activity. Therefore, when combined with recent literature, it is feasible to speculate that while nucleolar accumulation of mature/short extended hTR molecules upon TCAB1 inhibition blunts telomerase activity, the nucleolar accumulation of 5′-end monomethylated hTR precursors (after TGS1 inhibition) does not and even potentially increases the activity of this ribonucleoprotein. While the fate of hTR precursor molecules that are retained in the nucleolus remains to be determined, it is possible that this compartmentalization pattern contributes to the reduced MTR4 recognition and exosome decay observed after TGS1 inhibition, as ZCCHC8 and RBM7 (components of the NEXT complex) are excluded from the nucleolus (Lubas et al. 2011). These results corroborate the importance of cellular trafficking for telomerase biogenesis and further demonstrate how this compartmentalization can be regulated by posttranscriptional modifications to the 5′ and 3′ ends of hTR, potentially determining telomerase assembly or decay.

The posttranscriptional processing of hTR as a potential therapeutic target

Our data establish that sequential posttranscriptional modifications to the extreme ends of hTR molecules determine hTR maturation and decay. During the first step of processing, 5′-cap modification directly influences decay of long 3′ genomically extended hTR precursors by regulating exosome targeting. A second step of processing is then modulated by the 3′-end oligoadenylation of hTR, which controls the decay of mature and short 3′-end extended molecules. These sequential steps of regulation contribute to the tight control of telomerase activity in human cells, allowing for optimal tissue development and homeostasis. Impairment of these pathways leads to distinct but severe phenotypes in patients. Our work raises relevant new questions, such as identifying the molecular determinants underlying TGS1 specificity toward noncoding RNAs and elucidating additional nucleolar factors involved in hTR precursor processing as well as how cellular compartmentalization influences RNA stability and function. While these questions remain, it is clear that the elucidation of pathways that contribute to telomerase biogenesis can open novel avenues to be explored for regulation of telomerase activity. This further supports the use of RNA therapeutics as a potential approach for the clinical management of a large subset of TBD patients, targeting regulators of hTR processing (i.e., through antisense oligonucleotides, small interfering RNAs, or small molecules) that ultimately determine telomerase function in human cells.

Materials and methods

Cell culture, siRNA transfection, and chemical treatment regimen

Human fibroblasts were grown in Dulbecco's modified Eagle medium (DMEM; Gibco) supplemented with 15% FBS (Gibco) and 1% penicillin–streptomycin (Gibco) and maintained at 37°C in 5% CO₂ and 5% O₂. HeLa and HEK293T cell lines were cultured in DMEM supplemented with 10% FBS and 1% penicillin–streptomycin and maintained at 37°C in 5% CO₂. Cells were passaged at ∼80% confluency using 0.05% Trypsin-EDTA (Gibco).

For siRNA-mediated knockdown experiments, transfections were performed using Lipofectamine RNAiMAX (Invitrogen) at a final siRNA concentration of 30 nM (siRNA sequences are listed in Supplemental Table S4). Unless specified otherwise, cells were maintained under siRNA treatment for 9 days, with repeat transfections on days 3 and 6 to sustain efficient depletion. For EXOSC3, knockdown cells were transfected for 3 days only before harvest. For experiments utilizing chemical inhibition of TGS1 and PAPD5, cells were treated with DMSO, 30 µM sinefungin, 5 µM RG7834, or a combination of both compounds for 20 days and refreshed every 72 h.

Lentivirus production, cell infection, and generation of stable cell lines

Lentivirus was produced by transfecting HEK293T cells (5 × 10⁶ per 10 cm dish) with the following third-generation packaging plasmids: 3.5 µg of pMDL, 1.5 µg of pVSVG, 1.5 µg of pREV, and 7 µg of transfer plasmid using Lipofectamine 3000 (Thermo Fisher). Viral supernatants were harvested at 24 and 48 h, filtered (0.45 µm), and concentrated with Lenti-X concentrator (Takara). Concentrated virus was used fresh or stored at −80°C. HeLa cells were first transduced with rtTA lentivirus and selected with 500 µg/mL geneticin for 15 days. Stable rtTA cells were then transduced with Twin-strep:3xFLAG:TGS1 or Twin-strep:GFP lentivirus followed by dual selection with 500 µg/mL geneticin and 1 µg/mL puromycin for 8 days. H1-hESCs were transduced with shTGS1-expressing lentivirus (target: 5′-GCAACATTATAGTCAACTTTA-3′) and selected with 1 µg/mL puromycin in mTeSR1 media for 7 days. Knockdown efficiency was routinely monitored, and cells were maintained under selection. Plasmid details are in Supplemental Table S2.

RNA isolation, RT-qPCR, and Northern blotting

Total RNA was isolated using TRIzol reagent (Ambion) or the Direct-zol RNA Miniprep Plus kit (Zymo Research) following the manufacturer's protocols. Equal amounts of DNase-treated RNAs were reverse-transcribed using SuperScript IV (Invitrogen) in 10–20 µL reactions. RT-qPCR was performed using PowerUp SYBR Green master mix (Applied Biosystems) on a QuantStudio 6 Pro system and analyzed by the ΔΔC_t method. Primers are listed in Supplemental Table S1. For Northern blotting, RNA was mixed with gel loading buffer II (Invitrogen), heated for 5 min at 65°C, and separated on a 6% polyacrylamide/8 m urea/1× TBE gel. RNA was transferred to a Hybond-N⁺ membrane (GE) via semidry transfer (0.5× TBE, 1.5 mA/cm²) for 30 min and UV-cross-linked (254 nm, 1200 × 100 µJ/cm²). Membranes were prehybridized in ULTRAhyb buffer (Invitrogen) for 1 h at 42°C and then hybridized overnight with ³²P-labeled oligonucleotide probes (T4 PNK-labeled, purified via MicroSpin G-25 columns). Washes were done twice for 15 min each at 42°C with 2× SSC/0.1% SDS and once for 10 min at 42°C with 0.1× SSC/0.1% SDS. Signals were detected with a Typhoon phosphorimager (GE) and quantified using ImageJ.

Generation of inducible CRISPRi TGS1 HeLa cells

HeLa cells were sequentially transduced with lentiviruses carrying (1) rtTA (mCherry⁺ and G418⁺), (2) TRE4B-driven dCas9-KRAB-IRES-GFP (a gift from Eric Lander; Addgene plasmid 85556), and (3) a pLV[2gRNA]-TagBFP2-U6 vector encoding two sgRNAs targeting the TGS1. After rtTA transduction, cells were subjected to dual selection with 500 µg/mL geneticin followed by FACS enrichment of the top 10% mCherry⁺ cells. Subsequent dCas9-KRAB-IRES-GFP integration was verified by DOX-induced GFP expression and FACS-enriched GFP⁺ cells. Because all cells expressed BFP following sgRNA delivery, no further selection was required. The resulting pooled population (sgTGS1) displayed robust DOX-inducible repression of TGS1, validated by Western blot (antibodies utilized are listed in Supplemental Table S3). DOX⁻ cells from the same line served as controls in all assays. TRE-KRAB-dCas9-IRES-GFP was a gift from Eric Lander (Addgene plasmid 85556; http://n2t.net/addgene:85556; RRID: Addgene_85556).

Immunoprecipitation of endogenous TGS1

Cells were lysed in a buffer containing 50 mM Tris-Cl (pH 8.0), 150 mM NaCl, 2 mM MgCl₂, 0.5% NP-40, and 1 mM DTT, supplemented with a protease inhibitor tablet (cOmplete, EDTA-free, Roche) and 250 U/mL recombinant RNasin ribonuclease inhibitor (Promega). The lysates were cleared by centrifugation for 10 min at 4°C. Protein G Dynabeads (Invitrogen 10005D) were preincubated with either 4.5 µg of anti-TGS1 antibody (Bethyl Laboratories A300-815A) or rabbit IgG isotype control (Invitrogen 10500C) for 2 h and subsequently added to the lysates for overnight rotation at 4°C. Following this, the beads were washed five times with the lysis buffer and then split into two parts in a 1:3 ratio. SDS–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer was added to one part for immunoblotting. RNA was extracted from the other part, followed by reverse transcription and either semiquantitative PCR or quantitative PCR (qPCR). Semiquantitative PCR products were visualized on a 2.5% agarose gel. Fold enrichment was calculated using the percentage input method as follows: To normalize, the C_t value of the input was adjusted to represent 100% of the starting material using the formula adjusted input C_t = C_t(input) − log₂(100 ÷ percentage input used). The percentage input for each IP sample was then calculated using percentage input = 100 × 2^{[adjusted input C t− Ct(IP)]}.

Immunoprecipitation of TMG-capped RNA

TMG-capped RNA immunoprecipitation was performed as described previously with minor modifications (Hayes et al. 2018). Briefly, 50 µg of heat-denatured total RNA was resuspended in RNA binding buffer (50 mM Tris-Cl at pH 7.5, 150 mM NaCl, 1.5 mM MgCl₂, 0.1% NP-40, RNasin ribonuclease inhibitor [Promega]). Protein G Dynabeads (Invitrogen 10005D) were preincubated with either 4.5 µg of anti-2,2,7-trimethylguanosine antibody (Millipore Sigma MABE302) or mouse IgG isotype control (Invitrogen 10400C) for 2 h at 4°C and then added to the RNA sample for an additional 4 h of rotation at 4°C. After five washes with RNA binding buffer, RNA was extracted directly from the beads using TRIzol reagent (Ambion) followed by reverse transcription and either semiquantitative PCR or quantitative PCR (qPCR). Semiquantitative PCR products were visualized on a 2.5% agarose gel. For RNA quantification, the percentage input method was used as described above.

Transient transfection and immunoprecipitation of Twin-strep:3xFLAG:MTR4 and Twin-strep:3xFLAG:mCherry in HEK293T cells

HEK293T cells with DKC1 knockdown alone or combined with TGS1 or PAPD5 knockdown were transiently transfected with 1 pmol of Twin-strep:3xFLAG:MTR4 or mCherry plasmid using Lipofectamine 3000 (Thermo Fisher). After 48 h incubation at 37°C with 5% CO₂, cells were harvested for RNA immunoprecipitation (RNA-IP) and Western blot. Cells were lysed on ice in buffer (50 mM Tris-HCl at pH 8.0, 150 mM NaCl, 2 mM MgCl₂, 0.5% NP-40, RNase inhibitor, protease inhibitors). Lysates were cleared by centrifugation at 21,000g for 10 min at 4°C, and protein was quantified by BCA protein assay kit (Thermo Fisher Scientific). For IP, 0.5–1 mg of protein was incubated with 100 µL of Strep-Tactin XT 4Flow resin (IBA LifeSciences) prewashed with buffer W (1 m Tris-Cl, 1.5 m NaCl, 10 mM EDTA) and lysis buffer. Binding was performed for 2 h at 4°C followed by five washes in lysis buffer. Beads were processed for Western blot or RNA extraction (TRIzol, Invitrogen) followed by cDNA synthesis and qPCR. To quantify the enrichment of hTR (amplicons A–D) in MTR4 pull-down relative to mCherry pull-down, qPCR C_t values were normalized to GAPDH using the equation ΔC_t = C_thTR − C_tGAPDH. Enrichment was calculated using the 2^−ΔΔCt method, where ΔΔC_t was determined as ΔΔC_t = ΔC_tMTR4 − ΔC_tmCherry. The relative fold enrichment of hTR in MTR4 pull-down was expressed as enrichment = 2^−ΔΔCt. To evaluate the impact of dual knockdowns compared with DKC1 knockdown alone, we further analyzed the relative enrichment of hTR across conditions. Fold enrichment values from DKC1 and TGS1 dual knockdown and DKC1 and PAPD5 dual knockdown were compared with the DKC1 knockdown alone condition.

PAPD5 RNA immunoprecipitation (RNA-IP) assays

HEK293T cells were transiently transfected with 1 pmol of Twin-strep:3xFLAG:PAPD5 or mCherry plasmid using Lipofectamine 3000 (Thermo Fisher Scientific). After 24 h incubation at 37°C with 5% CO₂, cells were harvested for RNA immunoprecipitation (RNA-IP) and Western blot. Cells were lysed on ice in buffer (50 mM Tris-HCl at pH 7.5, 300 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 5% glycerol, 2 mM EDTA, RNase inhibitor, protease inhibitors). Lysates were incubated for 30 min on ice with gentle mixing and cleared by centrifugation at 21,000g for 10 min at 4°C. Cleared lysates were first immunoprecipitated using anti-FLAG magnetic agarose beads (Thermo Fisher) pre-equilibrated with lysis buffer. Binding was carried out for 2 h at 4°C, followed by five washes in lysis buffer. Bound complexes were eluted with 3× DYKDDDDK peptide in lysis buffer for 30 min at 4°C, and the eluates were subjected to a second purification step using 100 µL of Strep-Tactin XT 4Flow resin (IBA LifeSciences) prewashed with buffer W (1 m Tris-Cl, 1.5 m NaCl, 10 mM EDTA) and lysis buffer. After 2 h incubation at 4°C and five washes in lysis buffer, beads were processed for Western blot or RNA extraction using TRIzol (Invitrogen) followed by cDNA synthesis and qPCR. For RNA quantification, C_t values were first normalized to U1 RNA. The percentage input method was then applied as described above. Enrichment of hTR in PAPD5 pull-down was expressed as fold enrichment relative to the mCherry control.

Fluorescence in situ hybridization and immunofluorescence detection

Fluorescence in situ hybridization (FISH) was performed using a set of 16 independent Quasar 670-labeled Stellaris probes (Biosearch Technologies), complementary to nucleotides 452–1316 downstream from the annotated hTR 3′ end (probe sequences are in Supplemental Table S1. Briefly, cells grown on coverslips were washed once with 1× RNAse-free PBS and fixed in 4% paraformaldehyde for 15 min at room temperature. After fixation, cells were washed three times with 1× PBS, permeabilized with 70% ethanol, and incubated overnight at 4°C. The following day, cells were washed twice with wash buffer (2× SSC, 10% formamide) for 5 min at room temperature and incubated overnight at 30°C in a humidified, dark chamber with 50 µL of hybridization buffer (2× SSC, 10% deionized formamide, 20% dextran sulfate) containing 0.125 µM Quasar 670-labeled Stellaris probes and a rabbit monoclonal antifibrillarin antibody (1:100; Cell Signaling Technology 2639T). Subsequently, cells were washed three times for 5 min each at room temperature with wash buffer, incubated with Alexa 488-conjugated secondary antibody (1:200; Invitrogen A-11008) for 45 min at room temperature, and washed three times for 5 min with 1× PBS containing 0.05% Triton X-100. Coverslips were mounted with VectaShield containing DAPI (Vector Laboratories H-1200-10). Images were acquired using a Zeiss LSM 980 confocal microscope at 63× magnification and analyzed using ImageJ software.

RNA ligase-mediated 3′-RACE with deep sequencing

RNA ligase-mediated 3′-RACE was conducted following a previously described method (Moon et al. 2015). DNase-treated RNA (600 ng) was ligated to 5 µM 5′-adenylated, 3′-blocked adapter (universal miRNA cloning linker NEB S1315S, sequence: 5′-rAppCTGTAGGCACCATCAAT–NH2-3′) with 250 U of T4 RNA ligase-truncated KQ (NEB M0373S), 25% PEG 8000, and 1 µL of RNaseOUT (Invitrogen 10777019) in a 20 µL reaction for 16 h at 25°C. Ligated RNA was cleaned up with RNA Clean&Concentrator columns (Zymo Research R1014), and cDNA was synthesized with universal RT primer and SuperScript IV. PCR amplification was carried out using universal RT primer (sequence: 5′-CTACGTAACGATTGATGGTGCCTACAG-3′) and hTR_L3 (sequence: 5′-CTCTGTCAGCCGCGGGTCTCT C-3′) with Phusion high-fidelity DNA polymerase (NEB M0530). PCR products were subjected to QIAquick PCR purification columns (Qiagen) for library preparation using the TruSeq Nano DNA LT library preparation kit (Illumina). The libraries were submitted to the Tufts University Genomics Core for sequencing and data analysis. Data processing and analysis were performed as described previously. The script used is available on GitHub (https://github.com/alberttai/tailAcount).

Quantification and statistical analysis

Statistical analyses were conducted using GraphPad Prism software. Error bars indicate the standard error of the mean (SEM). A P-value of <0.05 was considered statistically significant. Additional statistical details are provided in the specific figure legends.

Data and code availability

All raw read data (FASTQ files) for 3′-RACE sequencing are publicly available at the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo) under accession number GSE293934 (reviewer token: cbyvwwoefhsblgd). Western blot, Northern blot, and microscopy data reported here will be shared by L.F.Z.B upon request. No original code was generated in the study. Any additional information required to reanalyze the data reported here is available from L.F.Z.B. upon request.

Competing interest statement

The authors declare no competing interests.