The chromatin remodelling factor ATRX suppresses R-loops in transcribed telomeric repeats

Abstract

ATRX is a chromatin remodelling factor found at a wide range of tandemly repeated sequences including telomeres (TTAGGG)_n ATRX mutations are found in nearly all tumours that maintain their telomeres via the alternative lengthening of telomere (ALT) pathway, and ATRX is known to suppress this pathway. Here, we show that recruitment of ATRX to telomeric repeats depends on repeat number, orientation and, critically, on repeat transcription. Importantly, the transcribed telomeric repeats form RNA-DNA hybrids (R-loops) whose abundance correlates with the recruitment of ATRX Here, we show loss of ATRX is also associated with increased R-loop formation. Our data suggest that the presence of ATRX at telomeres may have a central role in suppressing deleterious DNA secondary structures that form at transcribed telomeric repeats, and this may account for the increased DNA damage, stalling of replication and homology-directed repair previously observed upon loss of ATRX function.

Keywords: ATRX; G‐quadruplex; R‐loops; telomeres.

Figure EV1. Cell‐type variability in ATRX binding is associated with differential transcription

1. Overlap between ATRX binding sites in mouse embryonic stem cells (ES), embryonic fibroblasts (MEF) and erythroid foetal liver (FL) cells.

2. Overlap between ATRX binding sites at genes in isogenic MEF and FL cells.

3. Reads per kilobase of transcript per Million mapped reads (RPKM) of RNA‐seq data in MEF and FL cells at: all genes in the genome; genes which bind ATRX in FL cells but not MEFs; genes which bind ATRX in MEFs but not FL cells; genes which bind ATRX in both MEF and FL cells (shared targets). Boxes represent the 25^th, median and 75^th percentiles. Whiskers represent the 10^th and 90^th percentiles. Statistical significance was determined using a Mann–Whitney U‐test on the difference in expression between FL and MEFs (MEF‐FL) in each group compared to (MEF‐FL) values of all genes in the genome.

4. %GC content of tandem repeats underlying ATRX peaks which were transcription correlated or transcription independent in MEF and FL cells. Transcription‐independent ATRX TRs (n = 35) had average %GC content lower than transcription‐correlated (n = 169) ATRX TRs. Boxes represent the 25^th, median and 75^th percentile of %GC content. Statistical significance was assessed using a two‐tailed Mann–Whitney test.

5. %GC content of tandem repeats underlying ATRX peaks which were H3K9me3 correlated or H3K9me3 independent in MEF and ES cells. H3K9me3‐independent ATRX TRs (n = 83) had average %GC content higher than H3K9me3‐correlated (n = 622) ATRX TRs. Boxes represent the 25^th, median and 75^th percentile of %GC content. Statistical significance was assessed using a two‐tailed Mann–Whitney test (see also Table EV1).

Figure 1. ATRX recruitment to telomeric sequences is dependent on transcription and length of repeats

1. Diagram of the ectopic reporter gene GFP containing a G‐rich repeat, integrated into the genome of 293T‐Rex at a single FRT site by site‐directed recombination. The integration inactivates LacZ‐Zeo and activates hygromycin resistance, which allows selection of stable clones. The expression of the ectopic gene is regulated by a pCMV‐inducible promoter, which is activated by addition of doxycycline. The blue boxes are GFP exons; lines are introns; and zigzag lines are the G‐rich tandem repeat (see also Appendix Fig S1).

2. Diagram shows locations of the qPCR amplicons used to assess ATRX enrichment at the ectopic telomere repeat region. Numbers indicate distance to the start of the ectopic cassette. The diagram is not drawn to scale.

3. ATRX ChIP analysis in stable clones containing telomere repeats of indicated sizes and the control without the repeats when the ectopic gene is inactive (transcription off).

4. ATRX ChIP analysis in stable clones containing telomere repeats of indicated sizes (clones 5, 6 and 7 for (TTAGGG)₇₁ construct and clones 5 and 6 for (TTAGGG)₄₂ construct) and the control without the repeats (clone EV4 and N2S‐9) when the ectopic gene is activated (transcription on) upon addition of 1 μg/ml doxycycline for 24 h. The enrichment of ATRX is represented as % input normalised to that at 16p telomeric region (16ptel). DIST is a negative control, and ribosomal DNA (rDNA) is a positive control.

Data information: Data bars were plotted as the mean of two independent experiments for the No repeats clones (n = 2), of three independent experiments for the (TTAGGG)₄₂ clones (n = 3), and of four independent experiments for the (TTAGGG)₇₁ clones (n = 4). Error bars represent standard error of the mean (SEM) where n is equal or greater than 3. Statistical significance was determined using Student's unpaired t‐test.

Figure EV2. Expression of GFP exon 1 and exon 2 upon doxycycline induction

1. A, B

   Reverse transcription quantitative PCR showing that both GFP exon 1 and exon 2 in (A) the (TTAGGG)₇₁ clones with G‐rich strand the template and (B) the reverse orientation clones with G‐rich strand the non‐template are expressed upon transcription induction by doxycycline. Data are presented as fold change of expression relative to GAPDH. “No reverse transcriptase” (no RT) controls were included in each experiment. Data bars represent the mean of at least four biological repeats (± SEM). Statistical significance was determined using Student's t‐test.

Figure EV3. ATRX recruitment to the G‐rich ψζ VNTR is dependent on transcription and the length of repeats

1. Diagram shows locations of the qPCR amplicons used to assess ATRX enrichment at the ectopic telomere repeat region. Numbers indicate distance to the start of the ectopic cassette.

2. ATRX ChIP analysis in stable clones containing ψζ VNTR of indicated sizes and the control without the VNTR when the ectopic gene is inactive (transcription off). The enrichment of ATRX is represented as % input normalised to that at 16p telomeric region (16ptel).

3. ATRX ChIP analysis in stable clones containing ψζ VNTR of indicated sizes when the ectopic gene is switched on (transcription on) by addition of 1 μg/ml doxycycline for 48 h. The enrichment of ATRX is represented as % input normalised to that at 16p telomeric region (16ptel).

Data information: Data bars were plotted as the mean of two independent experiments for the 140 bp VNTR (n = 2) and 390 bp VNTR (n = 2), of three independent experiments for the NoVNTR (n = 3) and the 240 bp VNTR (n = 3), and of four independent experiments for the 490 bp VNTR (n = 4). Error bars represent standard error of the mean (SEM) where n is equal or greater than 3. DIST is a negative control, and ribosomal DNA (rDNA) is a positive control. Statistical significance was determined by unpaired Student's t‐test.

Figure 2. ATRX binding at the ectopic G‐rich sequence may be modulated by H3K4me3, but is independent of H3K36me3, H3K9me3 and nucleosome occupancy

1. Diagram shows locations of the qPCR probes used to assess enrichment of relevant histone modifications at the ectopic telomere repeat region. Numbers indicate distance to the start of the ectopic cassette. The diagram is not drawn to scale.

2. H3K4me3 ChIP in (TTAGGG)₇₁ clones when transcription is off (Unind) and on (Ind). GAPDH promoter (hGD‐pro) is a positive control, and HBA promoter (HBA‐pro) is a negative control. n = 3.

3. H3K36me3 ChIP in (TTAGGG)₇₁ clones when transcription is off (Unind) and on (Ind). hGD‐body (a region in intron 1 of human GAPDH) is a positive control, and HBA body (a region in HBA gene) is a negative control. n = 3.

4. H3K9me3 ChIP in (TTAGGG)₇₁ clones when transcription is off (Unind) and on (Ind). hGD‐pro is a negative control, and 16ptel is a positive control. n = 4.

5. Histone H3 ChIP in (TTAGGG)₇₁ clones when transcription is off (Unind) and on (Ind) compared with and IgG antibody control. HBA‐pro and 16ptel are positive controls, and hGD‐pro is a negative control. n = 3.

Data information: For each panel, error bars show standard error of the mean of at least three independent experiments of two clones 6 and 7. In (B–D), enrichment level of the relevant histone mark is expressed as % input and normalised to that of histone H3 (see also Table EV1).

Figure 3. R‐loop formation increases upon transcriptional induction of the ectopic telomere repeat

1. Diagram shows the ectopic locus and locations of the qPCR probes used to assess R‐loop signal and ATRX enrichment at the ectopic telomere repeat region.

2. DIP analysis in cells containing the (TTAGGG)₇₁ ectopic repeat when transcription is off (Unind) and on (Ind). As a control, DIP samples were also treated with E. coli recombinant RNase H for 5 h at 37°C prior to immunoprecipitation with S9.6 antibody. Ribosomal DNA is a positive control, and DIST and Znf180 are negative controls. DIP signal is expressed as % input normalised to the non‐repetitive region of GFP exon 1. Data bars represent the average value from four independent experiments (± SEM) from three independent clones (5, 6 and 7) for the ectopic repeat (TTAGGG)₇₁. Statistical significance was determined using Student's paired t‐test.

3. Diagram shows the inversion of the repeat region. The G‐rich strand in the original orientation construct is the non‐template strand, whereas it is the template strand in the reverse orientation construct.

4. ATRX ChIP analysis in the reverse orientation clone when transcription is inactive (Unind) and active (Ind) upon addition of 1 μg/ml doxycycline for 24 h. ATRX enrichment is presented as % input normalised to that of 16ptel region. Data bars are the average values of at least three independent experiments (± SEM) from two clones (3 and 4) of the reverse orientation and three clones (5, 6 and 7) of the original orientation. rDNA and DIST are a positive and negative control, respectively. Statistical significance was determined using Student's paired t‐test.

Figure EV4. R‐loop formation preferentially occurs when the G‐rich strand is the non‐template strand

DIP analysis in clones containing the ectopic telomeric sequence with the G‐rich strand being the non‐template strand and in clones containing the reverse oriented telomeric sequence with the G‐rich strand being the template strand after activation of transcription by doxycycline. DIP samples were also treated with E. coli RNase H prior to immunoprecipitation with S9.6 antibody as a control. DIP signal is expressed as % input normalised to the non‐repetitive region of GFP exon 1. Data bars represent the average value from four independent experiments ± SEM. Statistical significance was determined by unpaired Student's t‐test.

Figure 4. Modulating R‐loop levels leads to changes in ATRX binding

1. A

   DIP analysis at endogenous regions of ribosomal DNA and 16p subtel. n = 3.

2. B

   DIP analysis at the ectopic telomeric sequence (TTAGGG)₇₁, treated with 10 μM camptothecin for 30 min or with solvent DMSO, following induction of transcription by 1 μg/ml doxycycline for 24 h. n = 3.

3. C, D

   Representative blot (C) showing a decrease in R‐loops at endogenous telomeres in the camptothecin (CPT)‐treated cells compared to the control DMSO‐treated cells. Immunoprecipitated DNA, using S9.6 antibody, was slot blotted and probed with telomeric probes. Quantification is shown in (D) for three biological replicates. R‐loop levels are expressed as % input.

4. E

   ATRX enrichment at the ectopic telomeric sequence (TTAGGG)₇₁, in cells treated with 10 μM camptothecin for 30 min or with solvent DMSO control, following induction of transcription by 1 μg/ml doxycycline for 24 h. n = 3.

5. F, G

   Representative blot (F) showing a reduction in ATRX binding at endogenous telomeres in the CPT‐treated cells compared to the DMSO control. Immunoprecipitated DNA, using H300 antibody, was slot blotted and probed with telomeric probes. Quantitation is shown in (G) for three biological replicates. ATRX enrichment is normalised to input.

6. H

   ATRX ChIP analysis at endogenous regions of ribosomal DNA and 16p subtel. n = 3. Enrichment of R‐loops or ATRX is presented as % input.

Data information: Data bars are the average values from three independent experiments ± SEM. Statistical significance was determined using Student's paired t‐test.

Figure 5. ATRX modulates R‐loop formation

1. Western blot analysis of whole‐cell extract of cells containing the (TTAGGG)₇₁ ectopic repeat treated with lentiviral shRNA against ATRX (shATRX‐2) or with shRNA control (shctrl). Western blot membrane was probed with anti‐ATRX and anti‐alpha‐tubulin antibody.

2. DIP analysis in cells containing the (TTAGGG)₇₁ ectopic repeat treated with shRNA against ATRX or shRNA control, followed by transcription induction by addition of 1 μg/ml doxycycline for 24 h. The enrichment of R‐loops at the ectopic repeats was measured by qPCR. n = 3. Znf180 is a negative control.

3. DIP analysis showing increase in R‐loops at ribosomal DNA region in the ATRX knockdown cells. DIST is a negative control. n = 3 (see also Fig EV5).

4. Western blot analysis showing re‐expression of ATRX in U‐2 OS 22/3 cell line upon induction by doxycycline for 2 days.

5. R‐loops at endogenous telomeres in U‐2 OS 22/3 before and after 2 days of doxycycline‐induced ATRX re‐expression. DIP was performed with genomic DNA from these cell lines. As a control, the samples were treated with recombinant RNase H prior to the immunoprecipitation. Recovered DNA was slot blotted on a Zeta‐probe blotting membrane and then hybridised with ³²P‐labelled telomeric oligos.

6. Quantitation of the S9.6 signal relative to the inputs in (E). n = 3.

Data information: For all panels, data bars are the mean from three independent experiments ± SEM. Statistical significance was determined using Student's paired t‐test.

Figure EV5. ATRX modulates R‐loops

1. Western blot analysis of whole‐cell extract of cells with the ectopic telomeric sequence (TTAGGG)₇₁ treated with lentiviral shRNAs against ATRX (shATRX‐90 and shATRX‐91) or with shRNA control. The Western blot membrane was probed with anti‐ATRX and anti‐alpha‐tubulin antibody.

2. DIP analysis in clones treated with shRNA against ATRX or shRNA control, followed by transcription induction by addition of 1 μg/ml doxycycline for 24 h.

3. DIP analysis showing an increase in R‐loops at ribosomal DNA region in the ATRX knockdown cells. DIST is a negative control.

Data information: The enrichment of R‐loops at the ectopic repeats was measured by qPCR. Data bars are the mean values from two independent experiments. Znf180 and DIST are negative controls.

Figure 6. Model explaining the role of ATRX at R‐loop sites

G‐rich regions are prone to form R‐loops during transcription, and these promote G4 formation. In the wild type, ATRX is recruited to these structures where it resolves R‐loops or recruits other enzymes that degrade R‐loops leading to dissolution of G4. In the absence of ATRX, R‐loops and G4s persist, which may cause impaired gene expression, stalling of replication and DNA damage.