Telomeres regulate human telomerase levels through TRF2 acting outside telomeres

To test if TL of cells affected TRF2 occupancy at the TERT promoter, we first elongated telomeres in cancer cells under isogenic backgrounds (termed as Short Telomere, ST, or Long Telomere, LT cells hereon). We used HT1080 fibrosarcoma cells (HT1080-ST) or the TL-elongated version HT1080-LT (with constitutive expression of TERT and the RNA component TERC; previously characterized by others and us, see Methods) (Mukherjee et al., 2018b; Cristofari and Lingner, 2006). TL difference between HT1080-ST/LT was confirmed by flow cytometry (FACS, Figure 2A and qPCR-based method, Figure 2—figure supplement 1A, and telomerase activity in Figure 2—figure supplement 1B). Chromatin-immunoprecipitation (ChIP) of TRF2 gave significantly reduced occupancy of TRF2 on the TERT promoter in HT1080-LT compared to HT1080-ST cells (Figure 2B, ChIP-qPCR spanning 0–300 bp upstream of transcription start site, TSS).

TRF2-dependent H3K27 trimethylation (Mukherjee et al., 2018b), including at the TERT promoter (Sharma et al., 2021), was reported earlier by us: TRF2 silencing reduced the H3K27me3 mark at the TERT promoter along with an increase in TERT expression in the different cell lines tested (Sharma et al., 2021). Here, we asked if TRF2-dependent TERT regulation was TL-sensitive. With TL elongation, the H3K27me3 mark was significantly reduced in HT1080-LT relative to the ST cells (Figure 2C). As expected from reduced promoter H3K27 trimethylation, TERT transcription was enhanced by >threefolds in HT1080-LT relative to ST cells; 3'UTR-specific primers were used to distinguish from exogenously expressed TERT (Figure 2D).

To rule out any effect of exogenous TERT (HT1080-LT cells above) we sought to elongate telomeres in a TERT-independent way. Using a reported method, telomeres were elongated in MDA-MB-231 breast cancer cells with guanine-rich telomeric repeat (GTR) oligonucleotides characterized by others and us, see Methods; termed MDA-MB-231-ST/LT; TL difference quantified using FACS, Figure 2A, and qPCR-based method, Figure 2—figure supplement 1, and telomerase activity in ; Mukherjee et al., 2018b; Wright et al., 1996. MDA-MB-231-LT cells had lower TRF2 binding at the TERT promoter compared to MDA-MB-231-ST cells (Figure 2B). Accordingly, H3K27 trimethylation at the TERT promoter was reduced and TERT upregulated by ~threefolds in MDA-MB-231-LT compared to MDA-MB-231-ST cells (Figure 2C–D). Together, TL-sensitive regulation of TERT was consistent in both HT1080 and MDA-MB-231 cells on telomere elongation.

As an alternative model, instead of elongation, we shortened TL in HCT116 colorectal carcinoma cells with a higher average TL (~5.6 kb) compared to HT1080 (~4.5 kb) or MDA-MB-231 (~3 kb) cells (Mukherjee et al., 2018b; Cristofari and Lingner, 2006; Myung et al., 2004). Two different methods were used for TL shortening:

(a) TERT-independent, telomere-specific sgRNA guided CRISPR-Cas9 to trim telomeres (see Methods, scheme in Figure 2—figure supplement 1C) or (b) depletion of the TERC RNA (see Methods, TLs in HCT116-ST/LT quantified using FACS; Figure 2A, and qPCR based method, Figure 2—figure supplement 1, and telomerase activity in Figure 2—figure supplement 1). The ST cells in both TL-shortened models had higher TRF2 binding on the TERT promoter compared to the HCT116-LT cells (Figure 2B). Also, we found a higher repressor H3K27me3 mark at the TERT promoter, as expected from increased TRF2, in ST compared to its LT counterpart in both the HCT116 models (Figure 2C). As expected, an increase in the H3K27me3 repressor mark at the TERT promoter resulted in > twofold reduced TERT expression in ST relative to the HCT116-LT cells (Figure 2D).

Previously, enhanced H3K27 trimethylation from TRF2-dependent binding of the RE1- silencing factor (REST) and EZH2 (the catalytic component of the PRC2 repressor complex) was reported at the TERT promoter (Sharma et al., 2021). Here, we asked if the TL-sensitive H3K27 trimethylation resulted from altered REST/EZH2 binding at the TERT promoter. ChIP of REST showed enhanced occupancy at the TERT promoter in ST relative to HT1080-LT cells (Figure 3A). EZH2 binding was also higher at the TERT promoter in ST compared to HT1080-LT cells (Figure 3B). Similarly, in MDA-MB-231 cells, REST and EZH2 occupancy at the TERT promoter was lower in LT compared to ST cells (Figure 3A–B). Furthermore, an increase in REST and EZH2 binding at the TERT promoter in ST relative to LT cells was clear in the HCT116-ST/LT cells (Figure 3A–B).

To further test TL-sensitive TERT promoter activity, we used a reporter cassette comprising 1300 bp of the TERT promoter upstream of the Gaussia luciferase (Gaussia Luc) gene. The reporter was inserted at the CCR5 safe-harbour locus, 46 Mb away from the nearest telomere, using CRISPR in HEK293T cells (Figure 4A). TL in HEK293T cells was shortened using telomere-specific sgRNA guided CRISPR-Cas9 to trim telomeres Figure 4B, see Methods, scheme in Figure 2—figure supplement 1. Higher TRF2 binding at the inserted TERT promoter in ST compared to the HEK293-LT cells was clear (ChIP-qRT primers specific to the CCR5-TERT insert were used; Figure 4A and C). Consistent with this, H3K27me3 deposition was relatively high in ST cells (Figure 4D), and the TERT promoter-driven Gaussia Luc reporter activity was reduced in the ST relative to the LT cells (Figure 4E). Together, these show TL-dependent TERT regulation through non-telomeric TRF2 as a likely intrinsic mechanism.

We next studied the effect of time-dependent (temporal) TL elongation/shortening, i.e., TL increase followed by a decrease in a continuous way. For this, we made stable HT1080 cells with doxycycline-inducible TERT. Following induction of TERT, and telomerase activity (Day 0), we checked TL through days 6/8/10/16 and 24 (Figure 5—figure supplement 1A). Gradual increase in TL to > fivefold by Day 10 (relative to Day 0) was evident; beyond Day 10, despite dox induction (and relatively high telomerase activity) further TL elongation was not seen (Figure 5—figure supplement 1A–B; ++/-- denotes presence/absence of dox, respectively).

For detailed analysis, we focused on days 0,10, and 24 as representative time points; with discontinuation of dox after Day 10 (Dox-HT1080, Figure 5A). Following the withdrawal of dox after Day 10, TL returned to roughly within 1.6-fold of the initial (Day 0) state by Day 24 (Figure 5B, Figure 5—figure supplement 1A).

TRF2 binding at the TERT promoter decreased at Day 10 with TL elongation. Conversely, as TL shortened beyond Day 10, TRF2 occupancy on the TERT promoter was regained on Day 24; Figure 5C. Notably, this showed TRF2 binding at the TERT promoter to be temporally controlled as a function of TL. Second, this further showed that TERT promoter-bound TRF2 was non-telomeric because telomere-associated TRF2 would increase with TL elongation (and looping), instead of decreasing as seen here.

We then examined the occupancy of REST and EZH2 at the TERT promoter at days 0, 10, and 24: a decrease in occupancy of both REST and EZH2 on Day 10, relative to Day 0, followed by an increase on Day 24 was clear (Figure 5D and E).

Accordingly, H3K27me3 deposition at the TERT promoter was lower on Day 10 compared to Day 0, followed by an increase on Day 24 (Figure 5F). Taken together, clearly, REST and EZH2 binding, and H3K27me3 occupancy were as expected from a decrease/increase in TRF2 binding at the TERT promoter as TL increased and then receded.

Consistent with this, expression of the endogenous TERT (using 3'UTR specific qRT-PCR primers) increased by Day 10 from Day 0 levels, and was significantly reduced by Day 24 compared to the Day 10 levels, although not reduced to the Day 0 state (Figure 5G), supporting the role of TL elongation in inducing or TL shortening in suppressing TERT.

We considered a second temporal model using stable MDA-MB-231 cells with dox-inducible TERT to exclude the possibility of cell-type-specific effects. After dox induction, we followed TL elongation/shortening and telomerase activity (through days 4/8/10 and 14, Figure 6—figure supplement 1A–B, ++/-- denotes presence/absence of dox, respectively). Like HT1080 cells, here also TL elongated till Day 10 and beyond Day 10, the presence of dox showed no further elongation in TL (data not shown). Like the HT1080 cell model, dox was withdrawn at Day 10, however, in the case of MDA-MB-231 cells, TL shortened to roughly the initial (Day 0) state by Day 14, in a relatively short timeframe than what was noted for HT1080 cells (Figure 6—figure supplement 1A).

For detailed analysis, here we selected days 0 and 10 with dox, withdrawal of dox at Day 10 (consistent with the HT1080 cell model), and thereafter Day 14 (given the enhanced reduction in TL in MDA-MB-231 compared to HT1080 cells) as representative time points (Dox-MDA-MB-231; Figure 6A). TL elongated to ~twofold by Day 10; and receded to approximately the initial (Day 0) length by Day 14 after stopping dox at Day 10 (Figure 6B).

Decrease in TRF2 binding, and corresponding reduced REST, EZH2, and the H3K27me3 mark at the TERT promoter were clear on Day 10 relative to Day 0 (Figure 6C–F). Promoter TRF2, REST, and EZH2 occupancy, including H3K27 trimethylation, was regained in the subsequent time point of Day 14, relative to the Day 10 levels, as expected from TL shortening after Day 10 (Figure 6C–F). Accordingly, we noted endogenous TERT transcript followed TL elongation/shortening: increased progressively till Day 10, and subsequently reduced gradually at Day 14 (Figure 6G). Cell-type specific differences in the experimental models, for instance, relatively low regain of TRF2 occupancy at the final time day points, Day 14 in MDA-MB-231 compared to Day 24 HT1080 cells, other than relatively enhanced pace of shortening in MDA-MB-231 cells vis-à-vis HT1080 cells on dox withdrawal were interesting to note. Nevertheless, taken together, data from two independent models of temporally altered TL, further supported promoter TRF2-binding mediated TERT regulation to be TL-dependent.

We next sought to test TL-sensitive TRF2 binding at the TERT promoter in vivo. Xenograft tumours were grown with either HT1080-ST or HT1080-LT cells in NOD/SCID mice (n=5 each, for tumour size characterization see Methods, Figure 7A). Following harvesting, tumours were first characterized: TL was overall higher in all LT tumours compared to HT1080-ST tumours; increased TERT and telomerase activity was retained in HT1080-LT tumours relative to ST (Figure 7B and C; Figure 7—figure supplement 1A). TRF2-ChIP from tumour tissue showed HT1080-ST had significantly more TRF2 occupancy at the TERT promoter than HT1080-LT, consistent with results from cultured cells (Figures 7D and 2B). Correspondingly, H3K27me3 mark deposition was reduced in xenograft HT1080-LT tumours compared to HT1080-ST (Figure 7E), supporting that TL-dependent TERT regulation was retained in cells growing in vivo.

G-quadruplex (G4) DNA secondary structures (Bryan and Baumann, 2011), widely reported as gene regulatory motifs (Rawal et al., 2006; Huppert and Balasubramanian, 2007; Verma et al., 2008; Verma et al., 2009), were found in the TERT promoter (Palumbo et al., 2009; Lim et al., 2010). G4-dependent promoter TRF2 binding was also recently shown to regulate TERT (Sharma et al., 2021). Here, we asked, whether and how TL-dependent TERT regulation was affected by the promoter G4s. We used the TERT promoter-Gaussia Luc-reporter inserted at the CCR5 locus in HEK293T cells described above (Figure 4A). G>A mutations at –124 or –146 positions from the TSS of the TERT promoter are frequently found to be clinically associated with multiple cancers and reported to disrupt the G4s (Akincilar et al., 2016b; Huang et al., 2013; Horn et al., 2013; Killela et al., 2013; Kang et al., 2016; Li et al., 2016; Akincilar et al., 2016a). These mutations were introduced in the reporter individually and a pair of cell lines with either long or short TL were generated (Figure 8A–B).

Low promoter TRF2 binding and reduced H3K27 trimethylation in cells with either –124 or –146 G>A mutations within the TERT promoter-reporter inserted at the CCR5 locus was clear from earlier work (Sharma et al., 2021). Here, based on low TRF2 binding at the G4-disrupted TERT promoter, we reasoned that TL-dependence of TERT regulation would be affected. TRF2 binding at the TERT promoter with G4-disrupting mutations was not regained upon shortening of TL and was consistent in the case of both –124G>A and –146 G>A mutations (Figure 8C, left and right panels). This was in contrast to the increase in promoter TRF2 binding observed on TL shortening in the case of the unmutated promoter (Figure 4C), supporting the role of G4s in TERT promoter TRF2 binding and thereby in TL-dependent TERT regulation. As expected, the difference in H3K27me3 deposition and Gaussia Luc activity was insignificant in –124G>A or –146 G>A ST cells compared to the corresponding LT cells (Figure 8D and E).

TRF2 R17 residue is required for the repression of TERT expression

To understand the role of TRF2 post-translational modification(s) (PTM), if any, on TERT repression, we screened TRF2 mutants (Walker and Zhu, 2012). TRF2 mutants arginine methylation-deficient R17H (Mitchell et al., 2009) acetylation-deficient K176R, K190R (Rizzo et al., 2017), or phosphorylation-deficient T188N (Tanaka et al., 2005) were expressed in HT1080 cells in a background where endogenous TRF2 was silenced (Figure 9—figure supplement 1A). As expected, TRF2 silencing induced TERT, and re-expression of TRF2-wild type (WT) repressed TERT. Interestingly, R17H and K176R increased TERT, whereas T188N and K190R gave relatively moderate change (Figure 9—figure supplement 1B). Noting that the N-terminal domain of TRF2 (with the R17 residue) was shown to interact with G4 DNA (Bryan and Baumann, 2011; Pedroso et al., 2009), and also the core histone H3 (Konishi et al., 2016), we focused on understanding the role of TRF2-R17 at the TERT promoter.

Dox-inducible TRF2-R17H or TRF2-WT stable lines were made with HT1080, HCT116, and MDA-MB-231 cells. Induction of TRF2-R17H gave significant upregulation of TERT; whereas, as expected, TRF2-WT repressed TERT in all three cell lines (Figure 9A, Figure 9—figure supplement 1C); dox dose-dependent TRF2 induction and resultant TERT levels shown in Figure 9—figure supplement 1D. The increase in TERT with TRF2-R17H expression in contrast to its reduction with TRF2-WT, was also clear from TERT FISH (Figure 9B).

To understand loss of function, we looked closely at TRF2-R17H DNA binding at the TERT promoter. The binding of FLAG-tagged-TRF2-R17H on the TERT promoter was insignificant compared to the FLAG tagged-TRF2-WT in HT1080 cells (Figure 9C). As expected from the loss of TRF2-R17H binding, REST, EZH2 and H3K27me3 occupancy was significantly low on the TERT promoter for TRF2-R17H compared to TRF2-WT (Figure 9D–F).

For a deeper understanding of the role of TRF2 in H3K27 trimethylation, we used in vitro histone H3 methyltransferase assay. Purified histone H3 along with the reconstituted PRC2 repressor complex was analyzed in the presence/absence of recombinant TRF2-WT or TRF2-R17H for methyltransferase activity (Figure 9G, Figure 9—figure supplement 1E-F). H3K27 trimethylation in the presence of H3 and the PRC2 complex was first confirmed. Following this, treatment with purified TRF2-WT gave further increase in H3K27 trimethylation, whereas TRF2-R17H did not lead to any significant increase, relative to respective controls (Figure 9G). Interestingly, together these support a direct function of TRF2 in histone H3K27 trimethylation in presence of the PRC2 complex, where the R17 residue of TRF2 plays a necessary role.

To test the effect of TL on TERT expression in a physiological setting where telomeres elongate or shorten, we used fibroblast cells along with corresponding iPSCs. Primary foreskin (FS) fibroblasts were reprogrammed to pluripotent stem cells (iPSCs) (see Methods, scheme in Figure 10A, characterization data for iPSC in Figure 10B). TERT mRNA expression (Figure 10C), telomerase activity (Figure 10D) and relative telomere length (Figure 10E), were confirmed, and as expected were higher in the iPSCs compared to parent FS fibroblast cells. Chromatin binding of TRF2, REST, and EZH2 proteins and deposition of H3K27me3 mark were analysed in FS fibroblast and derived iPSCs (Figure 10F–I). TRF2 occupancy in the iPSCs with longer telomeres was found to be lower on the TERT promoter in comparison to parent fibroblasts (Figure 10F). Furthermore, occupancy of REST (Figure 10G) and the repressor complex protein, EZH2 (Figure 10H) was reduced in the pluripotent stem cells. Corresponding to the reduced levels of TRF2 occupancy and resultant lower binding of the REST/PRC2 complex on the TERT promoter in the iPSCs, the repressor mark, H3K27me3 was significantly lower relative to FS Fibroblast cells (Figure 10I).

Next, we asked if TL played a causal role in TERT expression. That is, whether TL change in iPSCs influenced TRF2 occupancy at the TERT promoter and its subsequent impact on TERT. We used the telomere-specific sgRNA-guided CRISPR-Cas9 to trim telomeres (as reported earlier) (Chen et al., 2013). This resulted in a population of iPSCs with short telomeres ('iPSC-ST';~50% reduction in TL; see Methods, Figure 11A and B). A comparison of TRF2 occupancy between iPSC-ST cells and unaltered iPSCs revealed a marked increase in TRF2 binding at the TERT promoter (~ threefold) in the telomere-shortened cells (Figure 11C). As expected from this, based on the above findings, we observed an increase (of 1.5-fold) in the deposition of the repressive histone mark H3K27me3 in iPSC-ST cells compared to unaltered iPSCs (Figure 11D). Further, consistent with these, TERT was significantly reduced in iPSC-ST cells relative to unaltered iPSCs (Figure 11E). Taken together, these demonstrate how TERT repression results of TL shortening through non-telomeric TRF2.

HT1080 (obtained from ATCC) and its derivative cell line HT1080-LT (received as a kind gift from Dr. J. Ligner's lab) were maintained in Minimum Eagle's Essential Medium (MEM) (Sigma-Aldrich) with 10% Fetal Bovine Serum (FBS) (Gibco). HCT116 TP53 null (gift from Prof. Bert Vogelstein's lab), HCT116 WT, MDA-MB- 231, and HEK293T (purchased from NCCS, Pune), CRISPR CCR5 locus insert cell lines; along with their derivative cells (short or long telomeres) were grown in Dulbecco's Modified Essential Medium High Glucose (DMEM-high Glucose) (Sigma-Aldrich) with 10% FBS. Doxycycline (dox) Inducible TRF2 cells in HT1080 and Dox-HT1080 were cultured in MEM and dox Inducible TRF2 HCT116 and MDA-MB-231 and Dox-MDA-MB-231 cells were cultured in DMEM-High Glucose, all supplemented with 10% tetracycline-free Fetal Bovine Serum (FBS) (Clontech and DSS Takara). Cells were trypsinized and subcultured in desired culture vessels when at 90% confluency. All cells were grown in 5% CO2, 95% relative humidity, and 37 °C culturing conditions. Authentication of cell lines was carried out using STR profiling, and periodic PCR testing verified that no mycoplasma contamination was present.

Reagents catalog no:

Doxycycline (dox) inducible TERT and TRF2 post-translationally-modified (PTM) lentiviral stable cell generation

Doxycycline (dox) inducible TERT construct and TRF2 WT and R17H PTM variant construct- both in pCW57.1 backbone were individually transfected with 3rd generation lentiviral packaging plasmids (pRRE, pREV, and pMD2.G) in 4:3:3:1 ratio in HEK293 Lenti-X cells for viral particle generation. 12 hr post-transfection, media change was given and incubated in general culture conditions for 72 hr. Viral particles were then collected by filtering the media with a 0.45 µM syringe filter. This viral particle filtrate was then added along with fresh media and Polybrene (5 µg/mL) to HT1080 and MDA-MB-231 cells (for ind. TERT construct) and HT1080, HCT116 WT cells, and MDA-MB-231 cells (for ind. TRF2 variant construct); incubated for 24 hr. Post 24 hr, media containing viral particles was safely discarded and fresh media was added. Next, puromycin selection was given to the cells until the transduced population of cells could be stably maintained.

The TRF2 post-translational modification variants were generated using site-directed mutagenesis by PCR method. The primers containing the required mutation (as documented in the table) were used to amplify the entire pCMV6 plasmid containing the TRF2-WT sequence. The PCR product was purified and transformed into DH5⍺ E. coli cells. Then the mutated plasmid constructs were isolated and confirmed by Sanger sequencing.

Tumour xenograft generation in NOD SCID mice was outsourced to Vivo Bio Tech. Ltd.

Telangana, India. 2.5 million cells of HT1080, and HT1080-LT each were injected subcutaneously in healthy male mice (4–6 weeks old) and monitored for tumour growth. Tumours were harvested by sacrificing mice when tumour growth reached an average volume of 600 mm³(± 100).

Telomere length was determined using two methods - RT-PCR-based and Flow Cytometry-based. For flow cytometry-based evaluation using Agilent DAKO kit, 2×10⁶ cells were divided into two fractions- for FITC-labelled Telomeric PNA probe (Agilent DAKO kit) and unlabelled control cells. The cell fractions were washed and further labelled with PI for whole DNA content as per the manufacturer's protocol. The fluorescence signal was acquired for at least 10,000 events on BD Accuri C6 flow cytometer. FCS files were then analyzed using FlowJo version 10.8.1 software. Fold change of telomere length relative to reference cell samples was calculated using Median Fluorescence signal intensity values of labelled and unlabelled sample preparations, according to the kit-provided formula.

For RT-PCR-based telomere length estimation, genomic DNA was isolated using the Wizard Genomic DNA Purification kit. At least two dilutions of genomic DNA in the range of 10 ng - 60 ng were prepared for each sample and qRT-PCR was set up for telomere-specific primers and single copy gene 36B4 specific primer from O'Callaghan et al., 2008 and Cawthon, 2002, respectively (Cawthon, 2002; O'Callaghan et al., 2008). Ct value of telomeric primer was compared to respective sample 36B4 and then fold change over the control sample was calculated.

The primers used are (5' – 3'):

ELISA TRAP was performed as described in Sharma et al., 2021. Briefly, 1×10⁶ cells were lysed using mild lysis buffer CHAPS supplemented with RNase inhibitor to prevent loss of enzymatic activity (due to TERC degradation) and PIC to prevent protein degradation. Protein concentration in the lysate was estimated and diluted to 0.5 µg/µL concentration. 2 µg of protein was then used to set up telomerase repeat amplification protocol (TRAP) PCR (Kim et al., 1994). This was performed using a PCR master mix provided with ROCHE TeloTAGGG Telomerase PCR ELISA kit which uses 3' biotinylated telomeric sequence repeat primers, P1-TS primers. Telomerase from the lysate extends the P1-TS primer,, which in the next step acts as a template for amplicon generation using P1-TS primer and (TTAGGG)6 repetitive sequences- P2. After this PCR, dioxygenin (dig) tagged telomere-specific detection probes are hybridized to denatured PCR products. These tagged biotinylated PCR products are then bound to streptavidin-coated plates and quantified using anti-dig-Peroxidase conjugated antibody, which produces coloured product upon metabolizing substrate, TMB.

Relevant Primary antibodies (TRF2 Novus Biologicals #NB110-57130, REST Sigma Aldrich #17–641, EZH2 Cell Signal Technology # 5246, H3K27me3 Abcam #ab6002, H3 Abcam # ab1791, FLAG Merck #F3165) and suitable IgG (from Millipore, for Isotype control or Mock) were used to perform the ChIP assays as per protocol reported in Mukherjee et al., 2018b. Briefly, 3×10⁶ cells were harvested and fixed with 1% formaldehyde for 10 min. Post washing with ice-cold PBS, the fixed cells were lysed in SDS-based lysis buffer with 2 X mammalian Protease Inhibitor Cocktail (mPIC) and subjected to chromatin fragmentation to an average size range of 200–300 bp using Bioruptor (Diagenode). 10% fragmented chromatin was processed as Input fraction via phenol-chloroform-isoamyl Alcohol (PCI) and Ethanol precipitation. Assay was performed with 3 µg of respective primary antibody incubated overnight on rotor at 4 °C followed by immune complex pulldown with salmon sperm DNA saturated Protein G Magnetic Dyna Beads. Subsequent washes with low salt, high salt and then LiCl buffers were performed and beads resuspended in TE (Tris-EDTA pH 8) for proteinase K treatment at 55 °C, 2 hr. Thereafter, PCI was used to phase-separate DNA in aqueous layer; separated fraction was incubated with equal volumes of isopropanol, 3 M sodium acetate and glycogen overnight at –20 °C. Pulldown DNA was then precipitated by centrifugation and pellet washed in 70% ethanol before air drying and resuspending in nuclease free water. Chromatin pulldown was validated by q-PCR analysis.

ChIP-q-PCR for TRF2, REST, and EZH2 chromatin pulldowns was performed with equal amount of DNA for each fraction, determined by Qubit HS DNA kit. Then, fold change was calculated over IgG (Mock) by using average Ct values. In case of histone ChIP assays too, equal amount of DNA from each histone ChIP and its respective total H3 were used in qPCR. Ct values for Histone mark and total H3 were normalised to 1% Input and thereafter fold change over total H3 was calculated.

Cells were seeded a day before transfection at 70% confluency in a six-well plate. 1.5 µg plasmid per well was transfected by complexing with FUGENE at 1:3 (DNA: Reagent) ratio following the manufacturer's protocol. 12 hr post-transfection media change was given to the cells and later harvested for mRNA and protein analysis after 36 hr of media change.

Total RNA isolation was performed using TRIzol Reagent (Invitrogen, Life Technologies) as per the manufacturer's instructions. RNA quantification and cDNA preparation was done using the Applied Biosciences kit. Quantitative real-time PCR using SYBR Green-based method (DSS TAKARA) was employed to estimate relative transcript levels of mRNAs with GAPDH as housekeeping control gene for internal normalization. Average fold change was calculated by difference in internally normalised threshold cycles (Ct) between test and control samples. All experiments were performed in three independent biological replicates; with technical triplicates of each primer pair and sample combination for q-PCR analysis.

In HT1080 ST/LT, Dox-HT1080, and Dox-MDA-MB-231 cells with overexpression of TERT and TERC (for HT1080-LT), primer pair specific for TERT mRNA 3'UTR was implemented to capture the status of endogenous TERT expression in addition to standard primers spanning exon 15/16 (full length) and exon 7/8 (active reverse transcriptase) forms. 3'UTR TERT primers were designed with help from UTRdb 2.0.

mRNA primers (5' – 3'):

Untreated or transfected cells were seeded on coverslip for this experiment and processed according to Stellaris RNA FISH protocol. Briefly, cells were fixed with fixation buffer for 10 min at RT. This was followed by PBS wash and permeabilization with 70% (v/v) ethanol. For hybridization, 125 nM of probe per 100 μl of Hybridization buffer was mixed and added onto parafilm surface in a humidified chamber. Coverslips were then placed on this drop with cells side down (post-wash in wash buffer A containing formamide), incubated in dark at 37 °C overnight. After incubation, the coverslips were transferred to fresh 1 ml Wash Buffer A for 30 min. This was followed by nuclear staining with DAPI in the same buffer, in the dark for 30 min maximum. Finally, the coverslip was washed in Wash Buffer B for 5 min and then mounted on the slide with anti-fade mounting reagent.

Post slide processing, fluorescence signal for Quasar 670 TERT RNA probe and DAPI were acquired on Leica SP8 confocal microscopy system. Implementing the default Rolling ball radius feature of open-source image analysis software Fiji ImageJ, background signal subtraction of each image frame was performed. Thereafter, signal from n=30 cell nucleus (manually marked) was quantified.

Adherent cells were seeded on coverslips at a confluency of ~70% before transfecting them with required TRF2 overexpression plasmids. At the harvesting time-point, cells were fixed using freshly prepared 4% Paraformaldehyde and incubated for 10 min at RT. Cells were permeabilised with 0.1% Triton X-100 (10 min at RT). The cells were treated with blocking solution (5% BSA in PBS+0.1% Tween 20 (PBST)) for 1 hr at RT. All the above steps were followed by three washes with ice-cold PBS for 5 min each. Post-blocking, cells were treated with anti-FLAG antibody (Merck #F3165, 1:1000 diluted in 5% PBST) and incubated overnight at 4 °C in a humid chamber. Post-incubation, cells were washed alternately with PBS three times for 5 min and probed with secondary Ab mouse Alexa Fluor 488 (A11059, 1:3000) for 2 hr at RT, in the dark. Cells were washed again with PBS three times and mounted with Prolong Gold anti-fade reagent with DAPI. Images were taken on Leica TCS-SP8 confocal microscope.

1.5×10⁶ HEK293T cells (~60% confluency) were transfected with 3 µg of TRF2- WT or TRF2-R17H overexpression pCMV6 plasmid. 24 hr post-transfection, cells were treated with G418 selection for 36–48 hr. Post this period of overexpression under selection pressure, the cells were harvested and resuspended in around 3–4 mL ice-cold TBS (150 mM NaCl, 50 mM Tris-Cl, pH7.4) with 2 X mammalian Protease Inhibitor Cocktail (mPIC). The cells were then sonicated to lyse open at 15 s ON/ 30 s OFF cycles for 5 min (until turbidity was reduced), at 4 °C. Cell debris was then separated by centrifuging at 13,000 rpm, 5 min, at 4 °C. M8823 M2 FLAG Dynabeads were prepared for incubation with the cell lysate as follows: 40 µL of the bead slurry was washed/ equilibrated twice in TBS by loading on 1.5 mL tube magnetic stand, kept on ice. This equilibrated bead solution is then added to the cell lysate and incubated overnight at 4 °C.

Post incubation of lysate with M2 FLAG bead for binding, the supernatant solution was removed and beads (bound to tube on magnetic stand) were washed based on step gradient purification. Two washes with 1 mL each of 150 mM, 300 mM, 500 mM, and 1 M NaCl containing TBS were given to remove non-specific proteins bound to beads. All the wash fractions along with the supernatant and purified bead-bound TRF2 fractions were then analysed by running a 10% SDS-PAGE and staining with Coomassie Brilliant blue.

1 µg of purified histone H3 N-terminal peptide (procured as a component of Active Motif kit Catalog No. 56100) was incubated overnight (12–14 hr), at 37 °C in the presence of 8 µM (working concentration) S-adenosyl-L-methionine, 0.6 µg purified PRC2 complex of proteins (cat no.SRP0381) and in the presence/absence of M2 FLAG bead-bound TRF2 protein (WT or R17H variants) in 25 µL of a histone assay buffer (component of ab113454 kit). After that, the protocol of ab115072 –Histone H3 (tri-methyl K27) quantification kit was followed. Briefly, 25 µL of supernatant of the reaction mix was pipetted, added to the anti-H3K27me3 antibody-coated strip wells with 25 µL of antibody buffer (provided with kit ab115072) and incubated at room temperature for 2 hr; the wells were covered properly by Parafilm M to prevent evaporation. Post incubation, the wells were washed thrice with kit-provided wash buffer. Diluted detection antibody (as directed by kit) was then added to the wells and incubated for 1 hr at room temperature on an orbital shaker at 100 rpm. Next, the wells were washed thrice after aspirating the antibody solution. The suggested amount of Colour Developer solution was added to the wells and incubated in the dark at room temperature for 2–10 min. The colour development (blue) was monitored across the wells of sample and standard solution. Once the colour in the standard well containing a higher concentration developed a medium blue colour, 50 μL of Stop Solution was added to all wells to stop the enzymatic reaction. Upon stopping the reaction, a yellow colour develops, and the absorbance reading at 450 nm was noted within 15 min on a microplate reader. Fold change in trimethylation over reference condition is determined by the following formula: F o l d c h a n g e o f T r i m e t h y l a t i o n − = ( ) T r e a t e d s a m p l e O D b l a n k O D ( ) t e s t e d − / ( ) U n t r e a t e d s a m p l e O D b l a n k O D ( ) c o n t r o l − \begin{document}$$\displaystyle \begin {array} {ll}\rm {Fold\, change\, of\, Tri-methylation= \left(Treated \left(tested\right)sample\, OD-blank\, OD \right)}/{\left(Untreated \left(control\right)sample\, OD-blank\, OD \right) }\end {array}$$\end{document}

Blank in case of experiment was the reaction mix without PRC2 complex and TRF2 protein forms.

Western blot analysis for checking the integrity of the purified bead-bound protein was also carried out. An equal volume of bead (5 µL) was loaded for both TRF2-WT and TRF2-R17H on a 10% SDS-PAGE and transferred on polyvinylidene difluoride (PVDF) membranes (Biorad). Membranes were blocked with 5% skimmed milk and then probed suitably with primary antibodies- anti-TRF2 antibody (Novus Biological), and anti-FLAG antibody (Merck). Anti-rabbit and anti-mouse HRP-conjugated secondary antibodies from CST, respectively were used. Blots were developed using BioRad ClarityMax HRP chemiluminescence detection kit and images acquired on the LAS500 GE ChemiDoc system.

Secreted Gaussia luciferase signal in the CRISPR-modified short telomere and control cells (WT, –124G>A and –146 G>A -ST/LT) was quantified using Promega Gaussia luciferase kit according to the manufacturer's protocol. The luminescence signal (arbitrary unit, a.u.) was normalised to total protein for each condition.

All experiments were performed with biological replicates, and data are presented as the mean ± standard deviation (SD) unless otherwise stated. Statistical significance was determined using appropriate tests based on experimental design.

For comparisons between groups where samples were derived independently (e.g. comparisons between each parental cell line and its corresponding short or long telomere-altered counterpart, or between control and TRF2-overexpressing conditions), an unpaired two-tailed Student's t-test was used. For experiments in which the same set of biological replicates was subjected to different treatments or time points (e.g. doxycycline induction and post-withdrawal), a paired two-tailed Student's t-test was applied.

For comparisons across more than two time points or conditions (e.g. telomere length analysis or expression dynamics during doxycycline treatment), one-way ANOVA followed by Tukey's Honestly Significant Difference (HSD) post-hoc test was used to identify statistically significant differences between groups.

For experiments involving two independent variables (e.g. Guassia luciferase assay across multiple conditions and time points, or in vitro histone methyltransferase activity under different treatments), two-way ANOVA was performed to assess the interaction effects. Where applicable, post-hoc comparisons were conducted to identify significant pairwise differences.

Statistical significance is indicated as follows: *p<0.05; **p<0.01; ***p<0.005; ****p<0.0001. All statistical analyses and data visualization were performed using GraphPad Prism version 8.0.2.