Super-resolution fluorescence imaging of telomeres reveals TRF2-dependent T-loop formation

Abstract

We have applied a super-resolution fluorescence imaging method, stochastic optical reconstruction microscopy (STORM), to visualize the structure of functional telomeres and telomeres rendered dysfunctional through removal of shelterin proteins. The STORM images showed that functional telomeres frequently exhibit a t-loop configuration. Conditional deletion of individual components of shelterin showed that TRF2 was required for the formation and/or maintenance of t-loops, whereas deletion of TRF1, Rap1, or the POT1 proteins (POT1a and POT1b) had no effect on the frequency of t-loop occurrence. Within the shelterin complex, TRF2 uniquely serves to protect telomeres from two pathways that are initiated on free DNA ends: classical nonhomologous end-joining (NHEJ) and ATM-dependent DNA damage signaling. The TRF2-dependent remodeling of telomeres into t-loop structures, which sequester the ends of chromosomes, can explain why NHEJ and the ATM signaling pathway are repressed when TRF2 is present.

Figure 1. STORM imaging of telomeres in intact MEFs and mouse splenocytes

(A) Comparison of conventional (Conv) and 3D-STORM (STORM) images of MEF telomeres detected by FISH. MEFs fixed on coverslips were hybridized with an Alexa 647-labeled [CCCTAA]₃ PNA probe. A conventional fluorescence image was taken (left), before the same area was imaged with 3D-STORM (right). The z-coordinates in the 3D STORM images shown here and in subsequent figures are color-coded according to the colored scale bar beneath the STORM image. The lower panels: zoom-in images of the boxed regions in the upper panels. (B) Enlarged area showing two telomeres imaged by 3D-STORM. (C) Distribution of the effective diameter of the telomere signals calculated as the diameter of a sphere of equivalent volume. (D) T-loop like architectures visualized in mouse splenocytes. Left: large field view showing several telomeres. Right: examples of individual telomeres. The images are from a small subset of nuclei that show telomeres in a relaxed configuration after cytocentrifugation.

Figure 2. STORM imaging revealing t-loops after chromatin spreading

(A) Schematic of the chromatin spreading procedure. (B) Conventional fluorescence image of a spread sample. A dense layer of decondensed string-like bulk DNA labeled with YOYO-1 (green) and FISH-labeled telomeres (FITC labeled [TTAGGG]₃ PNA probe, red) are visible. (C) 3D-STORM image of t-loops after the chromatin spreading procedure shown in (A) and (B). Linear, t-loop, and ambiguous “x” structures are classified according to the criteria described in the text and Fig. 3. Bottom, two enlarged t-loops. (D) Examples of t-loops detected as in (C). (E) Distributions of the total contour lengths of linear telomeric DNAs (n=224) and telomeric DNA exhibiting a t-loop configuration (n=58). (F) Distribution of the loop portion and the total contour length (loop + tail) of telomeric DNAs in a t-loop configuration (n=58). (G) Distribution of t-loops based on the relative size of the loop part as a fraction of the total contour length (n=58). See Fig. S1 for additional t-loop images.

Figure 3. The frequency of t-loop occurrence is diminished when telomeres fuse

(A) Immunoblot for TRF2 (doublet indicated by the line) in SV40LT TRF2^F/−Cre-ER^T1 cells treated with 0.5 mM 4-OH tamoxifen (4OHT) and harvested after 156 h (+Cre). *, non-specific bands (loading control). (B) Metaphase spreads showing telomere fusion before and after TRF2 removal from cells treated as in (A). Green: telomeric FISH with a FITC-conjuaged PNA [CCCTAA]₃ probe. Red: DNA stained with DAPI. (C) Genomic blot for telomeric DNA demonstrating telomere fusions (indicated) after deletion of TRF2. The gel image shows AluI/MboI-digested DNA hybridized with a TTAGGG repeat probe. AluI/MboI restriction enzymes digest non-telomeric genomic DNA and spare the telomeric TTAGGG repeats. (D) Representative STORM images before and after TRF2 deletion with Cre. T-loops, linear telomere structures (lin), ambiguous molecules (x) are indicated. (E) Examples of molecules scored as t-loops, linear telomeres, and ambiguous ‘x’ structures. (F) Percentage of the scored molecules (ambiguous ‘x’ structures excluded) that are in a t-loop configuration. Cells were treated as described in (A) and imaged as in (D). Graphs show mean and standard deviation (SD) values from 3 independent experiments (n≥200 molecules per experiment). P value from unpaired two-tailed Student's t-test. (G) Length distribution of linear telomeric DNAs detected by STORM imaging in TRF2^F/−Cre-ER^T1 cells before (-Cre; n=224) and 156 h after (+Cre; n=357) Cre treatment. See Fig. S2 for t-loop counts with and without exclusion of ‘x’ structures and the results obtained at 72 h after Cre.

Figure 4. TRF2 is required for the formation and/or maintenance of t-loops

(A) Immunoblot for TRF2 (line) in SV40LT TRF2^F/−ATM^−/−Cre-ER^T1 cells before (-Cre) and after (+Cre) 4OHT treatment (156 h). *, non-specific bands (loading control). (B) Representative metaphase spread of cells as in (A). (C) In-gel assay for the telomeric 3’ overhang (native) and total telomeric DNA analysis (denat). Note minimal telomere fusions and telomeric overhang loss (<30%) after TRF2 deletion from ATM^−/− cells. (D, F) Representative STORM images of telomeres in cells as in (A) or at 72 h after induction of Cre with 4-OHT (F). (E) Percentage of the scored molecules that are in a t-loop configuration. Cells as in (A) were imaged as in (D). Graphs show mean and SD values from 3 independent experiments (n≥200 molecules per experiment). P value was derived from unpaired two-tailed Student's t-test. See Fig. S3 for details of the t-loop counting with and without exclusion of ‘x’ structures. (G) T-loop fraction of the total scored telomeric molecules detected by STORM imaging as in (F). -Cre: n=286; +Cre (72 h): n=287. Graphs show means and SEMs.

Figure 5. Deletion of TRF2 does not affect crosslink efficiency in telomeric DNA

(A) Schematic of the experimental procedure. MEF nuclei either treated with psoralen/UV or not were digested with increasing amounts of MNase (agarose gel on left). DNA from the mono-, di, tri-, and tetra-nucleosomal MNase products was isolated and half of each sample was heat-denatured. Re-annealing of interstrand crosslinked DNAs regenerates double-stranded (ds) DNA while non-crosslinked DNAs remains single-stranded (ss). Bottom: The indicated samples from wt MEFs were separated on agarose gels and bands were visualized with Ethidium Bromide (EtBr), which preferentially stains dsDNA. Signals migrating at the position of dsDNA fragments were quantified with ImageJ and the percentage of signal remaining for each heat-denatured sample relative to its non-denatured control reflects the % crosslinking. (B) TRF2 deletion does not alter on the psoralen/UV crosslinking efficiency in bulk and telomeric DNA substantially. Top, middle: dinucleosomal MNase products were isolated and processed as in (A). EtBr signals for the region between 300 and 400 bp (marked by the line) were quantified with ImageJ and the heat resistant signal intensity (a measure for the crosslinking fraction) was calculated from comparison of the signals in the top and middle gels. Bottom: The gel containing the heat-denatured samples was dried and hybridized with a ³²P-labeled [AACCCT]₄ probe. Note that only the non-crosslinked DNAs will hybridize. Signal intensities were quantified with ImageJ and normalized to the EtBr value of the non-denatured samples in the top gel. The telomeric DNA signal intensity value for the TRF2^F/− sample not treated with Cre, and not treated with psoralen/UV was set to 100% and the values for the other samples were expressed relative to this value. The inferred % crosslinking of the dinucleosomal telomeric DNA band is given below the image.

Figure 6. TRF1 and Rap1 are not required for t-loop formation/maintenance

(A) Immunblot for loss of TRF1 (line) in the indicated cells treated Cre. *, non-specific band (loading control). (B) In-gel assay for the telomeric 3’ overhang (native) and total telomeric DNA (denatured) before and after deletion of TRF1. Numbers below the overhang gel indicate the relative signal intensity of the ssDNA overhang normalized to the total telomeric DNA. (C) Example of telomere imaging by STORM of cells lacking TRF1 (144 h post 4OHT). (D) T-loop frequency before and after deletion of TRF1. Data are presented as mean +/− SEM from 2 independent experiments. (E-H) As for (A-D) but with experiments performed on conditional Rap1 knockout cells. See Fig. S4 for t-loop counts of both experiments.

Figure 7. TRF2 is the main shelterin protein required for t-loop formation and/or maintenance

(A) Representative STORM image of telomeres after deletion of POT1b from the indicated cells (144 h post 4OHT) (B) T-loop frequencies before and after deletion of POT1b. Data are presented as mean +/− SEM from 2 independent experiments. (C, D) As for (A, B) but involving co-deletion of POT1a and POT1b from the indicated cells with Cre (144 h time point). (E) Representative STORM images of telomeres before and after removal of the shelterin complex through co-deletion of TRF1 and TRF2 (156 h post 4OHT). (F) T-loop frequencies before and after deletion of TRF1 and TRF2. Data are presented as mean +/− SEM from 2 independent experiments. See Fig. S4 for details of the t-loop counts in B, D, and F and the pertinent analysis of the protein and DNA in the cells. (G) Model explaining how TRF2-mediated t-loop formation/maintenance protects telomeres from MRN-initiated ATM signaling and Ku70/80-initiated NHEJ. When TRF2 is absent, telomeres are converted into the linear structure thereby allowing access to Ku70/80 and MRN. Ku70/80 and MRN are excluded from the telomere terminus when telomeres are in the t-loop configuration. While Ku70/80 and the MRN complex are also found in association with the shelterin complex, these Ku70/80 and MRN are not depicted here. The functions of these shelterin-associated Ku70/80 and MRN are not known.