Mammalian 5' C-rich telomeric overhangs are a mark of recombination-dependent telomere maintenance

Abstract

Recent evidence for 5'-cytosine (C)-rich overhangs at the telomeres of the nematode Caenorhabditis elegans provided the impetus to re-examine the end structure of mammalian telomeres. Two-dimensional (2D) gel electrophoresis, single telomere-length analysis (STELA), and strand-specific exonuclease assays revealed the presence of a 5'-C-rich overhang at the telomeres of human and mouse chromosomes. C-overhangs were prominent in G1/S arrested as well as terminally differentiated cells, indicating that they did not represent replication intermediates. C-rich overhangs were far more prevalent in tumor cells engaged in the alternative lengthening of telomeres (ALT) pathway of telomere maintenance, which relies on the homologous recombination (HR) machinery. Transient siRNA-based depletion of the HR-specific proteins RAD51, RAD52, and XRCC3 resulted in changes in C-overhang levels, implicating the involvement of 5'-C-overhangs in the HR-dependent pathway of telomere maintenance.

Figure 1. Cytosine-rich telomeric overhangs at human and mouse chromosome ends

(A) Schematic representation of 2D gel electrophoresis, whereby DNA is separated according to its molecular weight (1D) and conformation (2D). Native electrophoresis allows for visualization of DNA overhangs and extra-chromosomal single-stranded telomeric material (left panel), whereas denaturing conditions establish the presence of double-stranded telomeric DNA of linear or circular conformations (right panel). (B) Restriction digested genomic DNA from human KMST6, Saos2, U2OS, IMR90, HeLa and mouse (MEF) cell lines was electrophoresed in two dimensions and probed for DNA of G-rich (top panels) or C-rich (bottom panels) telomeric sequence under native conditions. The high molecular weight signals of U2OS and mouse telomeres have been indicated. The ratio of the levels of G- to C-overhangs is shown below as an indication of C-overhang enrichment in ALT and mouse cells. Signal intensities corresponding to each overhang were normalized against the signal for their corresponding denatured controls, and then compared to each other. (C) As in B, except that the gels were denatured before hybridization.

Figure 2. The C-rich overhang is terminal and 5’ in orientation

(A) Genomic DNA from Saos2 cells was treated with a 5’-3’ (RecJf) or a 3’–5’ (Exo I) exonuclease, followed by restriction digest. 1D image of nuclease treated DNAs (20 µg) loaded onto an agarose gel and stained with ethidium bromide serves as a loading control. (B) Lanes corresponding to each DNA sample from (A) were excised and electrophoresed in the second dimension and in-gel hybridizations with strand-specific ³²P-end labeled oligonucleotide probes were performed to demonstrate DNA of G-rich (top panels) or C-rich (bottom panels) telomeric sequence under native conditions. (C) As in (B), except that the gels in (A) were denatured prior to re-hybridization.

Figure 3. The telomeric C-rich overhang persists across the cell cycle and is present in terminally differentiated cells

(A) Saos2 human tumor cells were synchronized by the thymidine/aphidiocolin double block method and collected and processed for native (left panels) and denaturing (right panels) 2D gel analyses at 0, 4, 8, 12 hr post release from cell cycle arrest. Corresponding cell cycle phases are indicated on the left and the cell cycle distribution in % in the right panel. The ratio of native to denatured signal was normalized to the ratio in asynchronous cells and has been indicated in the left panels. Ethidium Bromide staining of the first dimension is shown for loading accuracy (bottom panel). (B) C2C12 mouse muscle cell precursors or myoblasts were induced to differentiate into muscle myotubes by introducing 2 % horse serum in place of fetal bovine serum in the growth medium. Genomic DNA (25 µg) either from myoblasts (top panels) or myotubes (bottom panels) was extracted, restriction digested and processed for 2D gel analysis under native and denaturing conditions. The high molecular weight telomeric signal has been indicated. Ethidium Bromide staining of the first dimension of each gel is shown for loading accuracy (bottom panel).

Figure 4. The telomeric C-overhang is at least partly of chromosomal origin, but maintains close association with the extra-chromosomal telomeric compartment in ALT cells

(A) Schematic representation of the arcs resulting from native 2D gel electrophoresis of telomeric DNA from Saos2 Hirt extracts. Native electrophoresis allows for visualization of a DNA overhang (hollow arrowheads) and extra-chromosomal single-stranded telomeric material, including nicked/gapped double stranded circles (black arrowheads), single-stranded circles and linear DNA fragments (gray arrowheads). (B) Digested and undigested DNAs from the supernatant (Sup) and the accompanying pellet fraction of Hirt lysate preparations from Saos2 cells were separated by standard agarose gel electrophoresis in the first dimension and stained with Ethidium Bromide to determine equal loading. The molecular weight (MW) marker on the left is the 1 kb DNA ladder from Invitrogen with a MW range of 500 bp to 12 kb. (C) Undigested (left panels), AluI/MboI digested (center panels) Hirt extract Sup fractions and AluI/MboI digested pellet fractions (right panels) were hybridized with probes specific for the telomeric C or G strand under native (upper two panels) or denaturing (lower two panels) conditions.

Figure 5. RAD51 knockdown increases the prevalence of the C- but not G-rich telomeric overhangs

(A) Immunoblot with antibodies against RAD51 from KMST6 whole cell extracts prepared 72 hrs post transfection with control or RAD51 specific siRNAs. Tubulin was used as a loading control. (B) Ethidium Bromide staining of the first dimension of the 2D gel electrophoresis. 20 µg of DNA was loaded per lane. (C) Each lane in (B, upper panel) was excised, electrophoresed in the second dimension and in-gel hybridized with ³²P-end labeled (TTAGGG)₄ oligonucleotide probe specific for the C-strands of telomeric DNA under native and denaturing conditions. Treatment with control and RAD51 specific siRNAs is indicated. (D) Each lane in (B, lower panel) was excised, electrophoresed in the second dimension and in-gel hybridized with ³²P-end labeled (CCCTAA)₅ oligonucleotide probe specific for the G-strands of telomeric DNA under native and denaturing conditions. Treatment with control and RAD51 specific siRNAs is indicated. (E) Quantification of three to four independent experiments of RAD51 targeting. The ratio between native and denatured signals is indicated in arbitrary units. The error bars represent standard deviation.

Figure 6. RAD52 knockdown increases the prevalence of C- but not G-rich telomeric overhangs

(A) Immunoblot with antibodies against RAD52 from KMST6 whole cell extracts prepared 72 hrs post transfection with control or RAD52 specific siRNAs. Tubulin was used as a loading control. (B) Ethidium Bromide staining of the first dimension of the 2D gel electrophoresis. 20 µg of DNA was loaded per lane. (C) Each lane in (B, upper panel) was excised, electrophoresed in the second dimension and in-gel hybridized with ³²P-end labeled (TTAGGG)₄ oligonucleotide probe specific for the C-strands of telomeric DNA under native and denaturing conditions. Treatment with control and RAD52 specific siRNAs is indicated. (D) Each lane in (B, lower panel) was excised, electrophoresed in the second dimension and in-gel hybridized with ³²P-end labeled (CCCTAA)₅ oligonucleotide probe specific for the G-strands of telomeric DNA under native and denaturing conditions. Treatment with control and RAD52 specific siRNAs is indicated. (E) Quantification of overhang levels upon RAD52 depletion shown as the ratio between native and denatured signals and indicated in arbitrary units. The error bars represent standard deviation of three independent experiments. (F) Native dot blot of the G-rich linear telomeric products of the phi29-dependent C-circle assay. 25–250 ng of restriction digested genomic DNA from control or RAD52 depleted KMST6 cells was used in each assay. The blot was hybridized with ³²P-end labeled (CCCTAA)₅ oligonucleotide probe. (G) Quantification of dot blot signal in (F) expressed in arbitrary units.

Figure 7. A model for the possible genesis and maintenance of 5’ C-overhangs at human telomeres that are prone to recombination

In the context of telomere dysfunction (i) inter-telomeric recombination (ii) may be initiated through 3’ G-overhang mediated invasion of neighboring chromosome ends, which are then used as a template for subsequent DNA synthesis. Alternatively, given sufficient length, a looped and partially protected conformation (iii) may be adopted leading to an XRCC3- and NBS1-dependent resolution event (Wang et al., 2004), thought to give rise to extra-chromosomal ds nicked t-circles (iv) and truncated telomeres (v). The precise mechanism of t-loop cleavage is unknown allowing for the possibility that the end structure of the resultant telomere may not necessarily contain the conventional 3’ G-overhang but rather a 5’ C-rich counterpart (v). The telomere bearing a 5’ overhang may be less stable than its G-rich equivalent since there is no known human protein that has an affinity specifically for ss C-rich telomeric DNA. If sufficiently long it may self-invade to form a small T-loop with an exposed ss displacement loop (vi), perhaps making it susceptible to nuclease attack. It is conceivable that this C-overhang mediated loop formation may be the precursor for C-circles. Participation of these in RCA (viii) with accompanying lagging strand synthesis could give rise to a product that is linear ds extra-chromosomal telomeric array (ix). There is precedent for an RCA-dependent mode of telomere maintenance at mitochondria of C. parapsilosis (Nosek et al., 2005). In this organism the RCA product bears a 5’ overhang (ix), which allows for invasion into and recombination with chromosome ends (x) endowing them with a 5’ telomeric overhang (v) (Nosek et al., 1995; Nosek et al., 2005; Tomaska et al., 2009). We suggest that this could occur at telomeres of human ALT cells, given the prevalence of 5’ C-overhangs in these cells. Since t-loop formation can be facilitated by 5’ overhang mediated self-invasion (vi) in C. parapsilosis (Nosek et al., 2005) and also in the worm (Raices et al., 2008), the existence of such a structure in human cells is conceivable. In the context of diminished RAD51 levels, G-overhangs of short dysfunctional telomeres that are subject to recombination may be at a risk of degradation via the action of unknown 3’–5’ exonucleases (xi) further contributing to C-overhang formation (v). This also holds true for XRCC3 and RAD52, likely because these are involved in recruitment of RAD51 to overhangs and stimulating its strand transferase activity, respectively. Depletion of RAD52 results in an increase in C-circle levels (Figure 6F, G), supporting the notion that C-circles are generated upstream of RAD52 action and are a byproduct of 5’ overhang mediated intratelomeric recombination and/or a substrate for RCA. The 2D gel schematic on the right shows the arcs representing the DNA species discussed.