TERRA recruitment of polycomb to telomeres is essential for histone trymethylation marks at telomeric heterochromatin

Abstract

TERRAs are long non-coding RNAs generated from the telomeres. Lack of TERRA knockout models has hampered understanding TERRAs' functions. We recently identified chromosome 20q as one of the main origins of human TERRAs, allowing us to generate the first 20q-TERRA knockout models and to demonstrate that TERRAs are essential for telomere length maintenance and protection. Here, we use ALT 20q-TERRA knockout cells to address a direct role of TERRAs in telomeric heterochromatin formation. We find that 20q-TERRAs are essential for the establishment of H3K9me3, H4K20me3, and H3K27me3 heterochromatin marks at telomeres. At the mechanistic level, we find that TERRAs bind to PRC2, responsible for catalyzing H3K27 tri-methylation, and that its localization to telomeres is TERRA-dependent. We further demonstrate that PRC2-dependent H3K27me3 at telomeres is required for the establishment of H3K9me3, H4K20me3, and HP1 binding at telomeres. Together, these findings demonstrate an important role for TERRAs in telomeric heterochromatin assembly.

Fig. 1

Deletion of the TERRA-20q locus markedly affects TERRA expression. a Scheme depicting the WT and the CRISPR-deleted allele for the 20q-TERRA locus located in the subtelomere of the chromosome 20, q-arm. The position of the gRNAs (E1 and S2) and the primers used to genotype the deletions are also shown. The black arrows represent the primers to amplify the CRISPR-deleted allele and the white arrows the ones to amplify the WT allele located inside the 20q-TERRA locus. b Ethidium bromide gels showing the WT and the CRISPR-deleted allele for the 20q-TERRA locus detected by PCR in a WT cellular pool and in different clones of the U2OS cells. c RNA from the a WT cellular pool, WT expanded clones (#1 and 2), or 20q-TERRA KO clones (#A7, E1, E6, E8, and H8) from the USOS cell line was isolated and used for TERRA detection by RNA dot-blot with a probe against the TERRA-UUAGGG-tract; 18S serves as loading control. (Graph) TERRA quantification normalized by 18S (mean values ± s.e.m., n = 3 biological replicates). d Representative confocal microscopy images of RNA-FISH against TERRA-UUAGGG-tract (green) in the U2OS WT clones (#1 and 2) and in the 20q-TERRA KO clones (#A7, E1, E6, E8, H8, and C4). Scale bar, 10 μm. (Graph) Quantification of the total spot intensity per nucleus normalized by nucleus area (mean values ± s.e.m., n = cells analyzed). One-way ANOVA with Dunnett’s post test was used for the statistical analysis (*p < 0.05, **p < 0.01, and ***p < 0.001)

Fig. 2

Deletion of the 20q-TERRA locus decreases telomere length. a Representative frequency graphs of telomere length distribution (a.u.) measured in the U2OS WT pool (black), in the WT clones (#1 and 2; dark gray), and in the 20q-TERRA KO clones (#A7, E1, E6, and C4; light gray). The mean telomere length and the number of telomeres analyzed is shown. The red lines are arbitrary lines placed in the exact same position in each frequency graph to visualize differences between samples. b Graph showing the quantification of the mean telomere length in the U2OS cells WT pool, the WT expanded clones (#1 and 2), and in the 20q-TERRA KO clones (#A7, E1, E6, and C4) by HT-Q-FISH (mean values ± s.e.m., n = technical replicates). c Graph showing the percentage of short telomeres in the same settings. Short telomeres are considered those in the 10% percentile of the total telomere length distribution (mean values ± s.e.m., n = technical replicates). One-way ANOVA with Dunnett’s post test was used for the statistical analysis (*p < 0.05, **p < 0.01, and ***p < 0.001)

Fig. 3

TERRAs are essential for the assembly of heterochromatic histone marks at the telomere, including H3K27me3. a ChIP-dot-blot of the H3K9m3, b H4K20m3, c H3K4m3, d H3K9ac, e H4K16ac, and f H3K27me3 histone marks for the U2OS cells WT pool, WT clones (#1 and 2), and from the 20q-TERRA KO clones (#A7, E1, E6, and C4), hybridized with a southern probe against the telomeric repeat. ChIP-dot-blot for IgG was used as a control. IgG ChIP-dot-blot shared by different antibodies shows different exposure times according to the best exposure time required for each antibody. DNA input signal is also shown. Below the ChIP-dot-blot for each mark is shown the quantification of the immunoprecipitated telomeric repeats normalized by the input for each individual sample (left graph) and for all WT clones vs. the 20q-TERRA KO clones (right graph) (mean values ± s.e.m., n = independent clone). Student’s t-test was used for the statistical analysis (*p < 0.05, **p < 0.01, and ***p < 0.001)

Fig. 4

TERRAs bind PRC2 and modulates its own and HP1 recruitment to telomeres. a A TERRA biotinylated RNA oligo (S) was incubated with nuclear extracts from U2OS cells and their association with Ezh2 and SUZ12 was detected by western blotting. A biotinylated control RNA oligo corresponding to the complementary sequence (AS) of the same length as the biotinylated TERRA (N₄₈) was used as control. Biotin pull-down in the absence of RNA oligo (no oligo) was included to monitor inespecific binding to the beads. b Representative images of the average number of colocalizations found on double immunostaining to TRF2 (green) and SUZ12 (red) in the U2OS WT and 20q-KO clones. Arrowheads indicate colocalization events. Scale bar, 10 μm. (Left graph) Quantification of the colocalization in each of the WT and 20q-TERRA KO clones (mean values ± s.e.m., n = number of cells) and (right graph) in all WT vs. the 20q-TERRA KO clones (mean values ± s.e.m., n = independent clone). c Telomeric ChIP-dot-blot of SUZ12 in U2OS cells infected with scramble or SUZ12 shRNA. IgG was used as a control. DNA input is also shown. (Graph) Quantification of the signal from the immunoprecipitated telomeric repeats normalized by the input (mean values ± s.e.m., n = technical triplicates). d Representative confocal STED super-resolution images showing the colocalization between TRF2 (in green) and SUZ12 (in red) in U2OS cells. Zoom: a colocalization event. Scale bar 5 μm. e Telomeric ChIP-dot-blot for HP1 in WT and 20q-TERRA KO clones. IgG was used as a control. DNA input signal is also shown. (Left graph) Quantification of the signal from the immunoprecipitated telomeric repeats normalized by the input for each individual sample and (right graph) for all WT vs. 20q-TERRA KO clones (mean values ± s.e.m., n = independent clone). Student’s t-test was used for statistical analysis (*p < 0.05 and **p < 0.01)

Fig. 5

The PRC2 complex is critical for the deposition of heterochromatic histone marks and HP1 at telomeres. a Upon infection of U2OS cells with shRNAs against EZH2 and SUZ12, total protein was obtained and used for western blot detection of EZH2 and SUZ12. Actin was used as loading control. b U2OS were treated with increasing concentrations of the EZH2 inhibitor EPZ-6438 for 4 days. Nuclear protein extracts were used for western blot detection of H3K27me3. Actin was used as loading control. (Graph) Quantification is shown. c Representative images of the average number of colocalizations for TRF2 (green) and SUZ12 (red) in U2OS cells infected with a scramble or with EZH2 or SUZ12 shRNAs (left panel) or treated with vehicle or EZH2 inhibitor. Arrowheads indicate colocalization events. Scale bar, 10 μm. Below the images is shown the quantification of (left graphs) the total nuclear SUZ12 upon shRNAs or EZH2 inhibitor and (right graphs) the colocalization between TRF2 and SUZ12 (mean values ± s.e.m., n = number of cells). d Telomeric ChIP-dot-blot of the H3K27m3, H3K9m3, and H4K20m3 for U2OS cells infected with scramble, EZH2, or SUZ12 shRNAs. IgG was used as a control. IgG ChIP-dot-blot shared by different antibodies shows different exposure times according to the best exposure time required for each antibody. DNA input signal is also shown. Quantification of the immunoprecipitated telomeric repeat signal normalized by the input for each individual sample is shown below (mean values ± s.e.m., n = technical replicates). Student’s t-test was used for the statistical analysis (*p < 0.05, **p < 0.01, and ***p < 0.001)

Fig. 6

Global HP1 and its localization at telomeres depend on PRC2 levels. a Representative images of the average number of colocalizations of RAP1 (green) and HP1 (red) in the U2OS infected with scramble, EZH2, or SUZ12 shRNA (left panel) or in U2OS treated with vehicle or with EZH2 inhibitor. Arrowheads indicate colocalization events. Scale bar, 10 μm. Quantification of (left graphs) total HP1 and (right graphs) the colocalization between RAP1 and HP1 (mean values ± s.e.m., n = number of cells). b Telomeric ChIP-dot-blot of the HP1 protein for the U2OS cells infected with scramble, EZH2, or SUZ12 shRNA. IgG was used as a control. DNA input signal is also shown. Quantification of the immunoprecipitated telomeric repeats normalized by the input is shown below (mean values ± s.e.m., n = technical replicates). Student’s t-test was used for the statistical analysis (*p < 0.05, **p < 0.01, and ***p < 0.001)

Fig. 7

Model of TERRAs as a master regulator of the heterochromatic status of the telomere. Diagram showing how TERRA recruits the PRC2 complex (EED, EZH2, SUZ12) and directs it to the telomere. Upon binding of PRC2 to the telomere, PRC2 catalyzes H3K27 methylation. This mark facilitates then the deposition of H3K9m3 and H4K20m3 and the recruitment and stabilization of HP1 protein at the telomere, important to maintain the heterochromatic status of the telomere. In the 20q-TERRA KO cells PRC2 is not recruited to the telomere, and the heterochomatic marks and HP1 protein are not taking place