Local enrichment of HP1alpha at telomeres alters their structure and regulation of telomere protection

Abstract

Enhanced telomere maintenance is evident in malignant cancers. While telomeres are thought to be inherently heterochromatic, detailed mechanisms of how epigenetic modifications impact telomere protection and structures are largely unknown in human cancers. Here we develop a molecular tethering approach to experimentally enrich heterochromatin protein HP1α specifically at telomeres. This results in increased deposition of H3K9me3 at cancer cell telomeres. Telomere extension by telomerase is attenuated, and damage-induced foci at telomeres are reduced, indicating augmentation of telomere stability. Super-resolution STORM imaging shows an unexpected increase in irregularity of telomeric structure. Telomere-tethered chromo shadow domain (CSD) mutant I165A of HP1α abrogates both the inhibition of telomere extension and the irregularity of telomeric structure, suggesting the involvement of at least one HP1α-ligand in mediating these effects. This work presents an approach to specifically manipulate the epigenetic status locally at telomeres to uncover insights into molecular mechanisms underlying telomere structural dynamics.

Fig. 1

Tethered HP1α at telomeres locally increases H3K9me3. a Schematic of HP1α fused to TRF1. HP1α consists of a chromo domain (CD), a hinge, and a chromo shadow domain (CSD); AA (amino acid). b−d Fluorescence imaging of UM-UC3 cells cotransfected with mCherry-tagged TRF2 and EGFP-TRF1, EGFP-HP1α, or EGFP-TRF1HP1α-fusion 48 h after transfection (n = 15–21 nuclei). b Representative images. mCherry shown as magenta in merged image. Scale bar: 10 µm. c Quantification of % EGFP area per nucleus ****p < 0.0001; n.s. (no significance). d Quantification of % telomere per nucleus with colocalization of EGFP and TRF2 (mCherry) **p = 0.0065; ****p < 0.0001. The high apparent colocalization of HP1α with TRF2 (within the HP1α group) is partly caused by random, coincidental overlaps with telomeres due to widespread HP1α spots; X−Y planes are projections of z-stacks. c, d Significance is assessed by one-way ANOVA and Dunnett’s multiple comparison test with 95% confidence level. Error bars represent standard error of the mean (s.e.m.). e Experimental set-up for ChIP to follow the localization of stably expressed TRF1HP1α in UM-UC3 after blasticidin selection (Bsd) at ~PD25. f Experimental groups are immunoprecipitated with the indicated antibodies, and hybridized on a dot blot with either telomere or control centromere (CENPB) probe (n = 3 independent replicates). Upon signal normalization to 10% input, g TRF1HP1α shows increased HP1α at telomeres compared to controls Vector only (Vonly), TRF1 and HP1α ****p < 0.0001; h TRF1HP1α shows decreased H3 at telomeres *p = 0.0133. Upon normalization to H3 signal, i TRF1HP1α shows increased H3K9me3 at telomeres per H3 *p = 0.0101 while j there is no significant change of TRF2 occupancy at telomeres. n.s. (no significance) g−j The values for three independent experiments (Supplementary Fig. 1) are used to calculate the s.e.m. for each group. p values are calculated by two-tailed unpaired t test with 95% confidence level

Fig. 2

Telomere-tethered HP1α attenuates telomere extension by telomerase but does not accelerate replicative senescence. a Experimental set-up to study the impact of HP1α on telomerase-based telomere extension in UM-UC3. First infection: EGFP-tagged Vonly, TRF1, HP1α, or TRF1HP1α. b Telomere length analysis of TRF1HP1α, various controls, and untreated parental cells (Prn) with WT hTR overexpression from PD0 to ~PD30. c Quantification (average telomere length) shows TRF1HP1α attenuates the WT hTR overexpression-induced telomere extension. Similar findings are observed in two independent replicates. d Qualitative β-gal staining of BJ fibroblasts with earlier versus later PD. Bar: 100 µm. e Quantifications of relative β-gal fluorescence units are normalized to µg of protein. BJ PD68 shows significantly more β-gal fluorescence than BJ PD34 ****p < 0.0001. Two independent experiments; each contains triplicates. Error bars represent s.e.m. p values are calculated by two-tailed unpaired t test with 95% confidence level. f Experimental set-up to determine if TRF1HP1α accelerates replicative senescence. These analyses were performed only 10–12 days after infection, and during that period (~5–6 PDs) telomere shortening was minimal. Thus, it is unlikely that the lack of any effect on β-gal was due to adaptive compensation by other proteins or selection of cell subpopulations. Fibroblasts g BJ (PD67-70) or h WI-38 (PD44) show no significant difference in β-gal signal. BJ, three independent experiments each contain triplicates. WI-38, single experiment with triple replicates. Error bars represent s.e.m.

Fig. 3

Ligand binding function of HP1α CSD controls telomere extension. a Schematic diagram of mutations in HP1α fused to TRF1 (AA—amino acid). CD mutant V22M; CSD mutants I165A and W174A; N-terminal phosphorylation deficient mutant NS2A; hinge mutant KRKAAA. b Transient cotransfection of mCherry-tagged TRF2 (magenta in merged image) and various EGFP-tagged TRF1HP1α mutants respectively in UM-UC3 cells imaged after 48 h. Scale bar: 10 µm. ~20 nuclei were counted per group in c and d. c Quantification of % EGFP area per nucleus ****p < 0.0001; *p = 0.0260; n.s. (no significance). Significance is assessed by one-way ANOVA and Dunnett’s multiple comparison test with 95% confidence level. d Quantification of % telomeres per nucleus with colocalization of EGFP and TRF2 (mCherry). Consistently, V22M and V22MI165A show fewer total fusion protein spots per nucleus (Supplementary Fig. 4) because V22M lacks the ability to bind to other, widespread genomic regions. Thus, the slight reduction of % colocalization of V22M and V22MI165A with TRF2 is likely to be at least partially because of fewer random overlaps of telomeres with widespread HP1α spots. e Quantification of TRF2 foci; n = ~20 nuclei per group. c−e Error bars represent s.e.m. f, g Telomere length analyses of TRF1HP1α, WT or HP1α mutant variants with WT hTR overexpression across PD0 to ~PD30. h, i Quantifications (average telomere length) show CSD mutants I165A, W174A or double mutant V22MI165A revert the telomere extension attenuation phenotype of TRF1HP1α. Similar findings were observed in two independent experiments

Fig. 4

TRF1HP1α results in reduced TIFs induced by mutant hTR expression. Cells stably expressing TRF1HP1α (WT or mutant variants of HP1α) were infected with lentivirus containing WT hTR, or mutant hTR (47A or TSQ1) on day 0, selected for stable expression after 48 h, and analyzed on day 5. a Fluorescence microscopy images of representative cells expressing mutant hTR 47A stained with telomeric (Tel-Cy3) peptide nucleic acid (PNA) probes (magenta in merged image) via fluorescent in situ hybridization (FISH), antibody against DNA-damage repair protein marker 53BP1 (green in merged image), and counterstained with DAPI. Zoom-in images (the last row) correspond to yellow-squared regions of the row above. b % TIF per telomere of each nucleus is quantified; n = 94–159 nuclei combining data of three independent experiments. ****p < 0.0001; n.s. (no significance). a, b Scale bar: 10 µm. c Upon TSQ1 expression, TRF1HP1α results in fewer TIFs compared to Vonly, TRF1 or HP1α controls. TRFHP1α ~11.4% shows decreased TIFs compared to Vonly: ~19.7% ***p = 0.0008; TRF1: ~20.4% ***p = 0.0005; HP1α: ~17.9% *p = 0.0137 (n = 30–38 nuclei). b, c Significance is assessed by one-way ANOVA and Dunnett’s multiple comparison test with 95% confidence level. d Fluorescence images of control cells overexpressing WT hTR (n = 27–36 nuclei). Same color scheme as a. TIFs quantification in the presence of e WT hTR or f Vonly (n = 28–36 nuclei) show minimal baseline DNA damage at telomeres. b, c; e, f Error bars represent s.e.m.

Fig. 5

TRF1HP1α allele-specific protection effects upon si-TRF2-induced telomeric damage. 72 h after transfection, a TRF2 knockdown efficiency with antibody against TRF2 (anti-TRF2) and GAPDH (anti-GAPDH) as loading control. (−) si-non-targeting; (+) si-TRF2. Quantification of TIFs in b si-non-targeting (n = 32–47 nuclei per group) or c si-TRF2. Left *p = 0.0188, right *p = 0.0192, **p = 0.0042, ****p < 0.0001 (n = 31–48 nuclei per group). b, c Significance is assessed by one-way ANOVA and Dunnett’s multiple comparison test with 95% confidence level. Error bars represent s.e.m. Note the similar pattern among TRF1HP1α alleles in c compared to the corresponding allele pattern in Fig. 4b

Fig. 6

TRF1HP1α increases the fraction of irregular-shaped telomeres analyzed by STORM. a Similar telomere length (average and length distribution) across TRF1HP1α, I165A and control groups (TRF1, HP1α) at the time of analysis. b Top: Widefield conventional fluorescence image of representative UM-UC3 nucleus hybridized with Cy5-end-labeled C-strand telomeric PNA FISH probe. Images acquired contain ~35,000 frames with a z-depth-range of ~700 nm. Middle: the corresponding STORM image. Bottom: Overlay of conventional and STORM images. Bar: 5 µm. c Top: Representative reconstructed single telomere STORM images of TRF1HP1α and each corresponding Rg (nm) across a gradient. Bottom: Corresponding raw images of individual signal localization spots (displayed as dots) prior to image processing and reconstruction. Bar: 100 nm. d−g Rg of individual telomeres (dots) in 19 nuclei analyzed for each group d TRF1, e HP1α, f TRF1HP1α or g TRF1HP1αI165A. Y-axis: Rg (nm). X-axis: nucleus index. Each individual nucleus is distinguished by a different color. Each dot corresponds to one telomere. h Distribution of Rg (nm) represents as a violin plot showing frequency (width of density plot), median (white dot), interquartile range (bar), and 95% confidence interval (line). TRF1 (n = 38 nuclei, 437 telomeres), HP1α (n = 19 nuclei, 264 telomeres), TRF1HP1α (n = 47 nuclei, 552 telomeres), and TRF1HP1αI165A (n = 27 nuclei, 451 telomeres). Means of Rg are compared using ANOVA Tukey’s multiple comparisons with 95% confidence level ****p < 0.0001; left ***p = 0.0003; right ***p = 0.0001; n.s. (no significance). Mean of TRF1 Rg (84 nm) indicated as cut-off (dashed line) and i fractions of Rg equal or greater than the 84 nm cut-off in experimental groups

Fig. 7

Model for how enhanced heterochromatin by telomere-tethered HP1α impacts telomere maintenance. Diagram of working model. Bar: 100 nm. Below the drawings: Representative reconstructed single telomere STORM images. See text for details