Strand-Specific Postreplicative Processing of Mammalian Telomeres

See allHide authors and affiliations

Science  28 Sep 2001:
Vol. 293, Issue 5539, pp. 2462-2465
DOI: 10.1126/science.1062560


Telomeres are specialized nucleoprotein structures that stabilize the ends of linear eukaryotic chromosomes. In mammalian cells, abrogation of telomeric repeat binding factor TRF2 or DNA-dependent protein kinase (DNA-PK) activity causes end-to-end chromosomal fusion, thus establishing an essential role for these proteins in telomere function. Here we show that TRF2-mediated end-capping occurs after telomere replication. The postreplicative requirement for TRF2 and DNA-PKcs, the catalytic subunit of DNA-PK, is confined to only half of the telomeres, namely, those that were produced by leading-strand DNA synthesis. These results demonstrate a crucial difference in postreplicative processing of telomeres that is linked to their mode of replication.

Telomeres are nucleoprotein structures at the ends of chromosomes that are composed of repetitive G-rich sequence (TTAGGG in vertebrates) and a variety of associated telomeric binding proteins. Together, they form a dynamic terminal structure that “caps” the natural ends of linear chromosomes (1, 2). This cap prevents degradation of chromosome ends and protects against inappropriate recombination. TRF2 (3, 4) and the three subunits of DNA-PK—Ku70, Ku80, and the catalytic subunit DNA-PKcs (5–8)—are among the proteins that participate directly in capping mammalian chromosomes. Direct visualization of mammalian telomeres by electron microscopy has revealed the existence of terminal structures known as t loops (9), which are created when a telomere end loops back on itself and invades an interior segment of duplex telomeric DNA. By sequestering natural chromosome ends, t loops may render telomeres nonrecombinogenic. It has been proposed that formation of t loops is mediated by TRF1 and TRF2 and requires a single-stranded extension of the TTAGGG sequence (10). However, other mechanisms of end-capping cannot be formally excluded, particularly in light of the controversy over whether all telomeres have 3′ overhangs suitable for t loop formation (11, 12).

To investigate the capping mechanism, we used a dominant-negative mutant of TRF2, TRF2ΔBΔM, that lacks both the NH2-terminal basic domain and the COOH-terminal Myb domain. TRF2ΔBΔM removes endogenous TRF2 from telomeres, resulting in diminished 3′ overhangs, induction of end-to-end fusions, formation of anaphase bridges, activation of DNA damage checkpoints, and impaired cell growth (3, 4). The uncapped telomeres are presumably “repaired” by nonhomologous end-joining (NHEJ), resulting in covalent end-to-end ligations that preserve telomeric DNA at the point of fusion.

We expressed TRF2ΔBΔM for 5 days in two independent clones of HTC75 human fibrosarcoma cells (13). Microscopic examination revealed that 44 of 154 mitotic cells exhibited end-to-end chromosomal fusions (14). In those cells containing fusions, the average frequency was 3.1 per cell. All were chromatid-type fusions involving at least two (and frequently multiple) chromosomes joined together end to end. This type of aberration, which we designate as telomeric chromatid concatenates (TCCs), is noteworthy because it demonstrates that this type of telomeric fusion can form only after replication of telomeric DNA (fusions in G1 produce chromosome-type aberrations). The exclusive appearance of TCCs suggests that telomere replication is a prerequisite for fusion; i.e., TRF2-mediated end-capping follows telomere replication. The absence of chromosome-type telomeric fusions also indicates that cells with TCCs do not progress through a second cell cycle. This is perhaps not surprising because anaphase bridging—the inevitable result of the numerous dicentric chromosomes created by telomeric fusion—would be expected to impede cell division.

These observations focused our attention on the role of DNA replication in end-capping. Telomeres face special challenges during replication. The protective terminal structure of the telomere must not only disassemble in order to replicate, it must also regenerate after replication. The two telomeres at the end of each mitotic chromosome arm replicate from a single parental telomere. One is produced through leading-strand DNA synthesis, the other through lagging-strand synthesis, hereafter referred to as the leading-strand and the lagging-strand telomeres, respectively. Immediately after replication, leading-strand telomeres are blunt ended and lagging-strand telomeres have a short 3′ G-rich single-stranded overhang. Both types of ends may be processed further, perhaps by C-rich strand degradation and/or sequence addition by telomerase (11, 15,16). Because sister telomeres remain in close proximity during interphase, one might expect that impaired end-capping would lead to a preponderance of telomeric fusions between sister chromatids. However, out of the 135 fusions observed, none resulted from sister union. This finding is noteworthy because it indicates that, of the two telomeres replicated from the same template, only one acquires the ability to fuse to other telomeres in the presence of TRF2ΔBΔM.

To differentiate leading-strand from lagging-strand telomeres (Fig. 1), we used a procedure based on the strand-specific in situ hybridization technique of chromosome-orientation fluorescence in situ hybridization (CO-FISH) (17). Unique hybridization patterns are produced for each of the three possible types of chromatid telomeric fusions: leading-to-leading strand, leading-to-lagging strand, and lagging-to-lagging strand (Fig. 1D). Using this strategy, we sought to determine whether impaired end-capping is limited to telomeres synthesized by one mode of replication or, alternatively, whether fusion is random but exclusive. In the latter case, either a leading-strand or a lagging-strand telomere might engage in fusion, thereby preventing its sister telomere from doing the same. These experiments (18) revealed that TCCs were overwhelmingly the products of fusion between leading-strand telomeres (Fig. 2). If mere chance dictates the type of fusion, the expected ratio of fusion products depicted in Fig. 2 would be 1:2:1. Instead, 133 out of 135 observed fusions had a pattern consistent with fusion between two leading-strand telomeres, and none were of the leading-to-lagging strand type (Table 1). Two fusions were tentatively identified as lagging-to-lagging strand types, because they lacked hybridization signal at the point of fusion. However, because they also lacked one or more hybridization signals at other telomere sites, these could not be definitively classified. When analysis was restricted to the 106 fusion events displaying a full complement of hybridization signals, all were the result of leading-to-leading strand fusion. For a random process, only one-fourth of all fusions should be of this type; therefore, the difference from expectation is highly significant (χ2 test, P ≪ 0.001).

Figure 1

Identification of leading-strand and lagging-strand telomeres. (A) Cells expressing TRF2ΔBΔM are allowed to replicate their DNA once in the presence of bromodeoxyuridine (BU) and bromodeoxycytidine (BC) and are then collected in mitosis. Because DNA synthesis is semiconservative, opposite strands of the telomere sequence are bromo-substituted. Each mitotic chromosome has one parental DNA strand and one newly synthesized bromo-substituted strand. After the cells are fixed and dropped onto microscope slides, the bromo-substituted strands are degraded by sequential UV and exonuclease treatments. (B) Each sister chromatid of a mitotic chromosome now contains just one of the parental DNA strands. A labeled (TTAGGG)7 single-stranded probe hybridizes to those telomeres that were replicated by leading-strand synthesis. Likewise, a (CCC TAA)7 probe would hybridize and identify lagging-strand telomeres. (C) Viewed by fluorescence microscopy, mitotic chromosomes have two telomere signals (red) in contrast to the four signals observed with ordinary FISH. (D) Each of the three different types of TCCs can be identified by its unique hybridization pattern. Shown here are the patterns expected for the (TTAGGG)7probe.

Figure 2

Images of partial metaphases with several TCCs as detected by CO-FISH. Telomeric DNA is detected with Cy3 (red), and chromosomal DNA is counterstained with DAPI (blue). (Ato D) Fusion points of several concatenates in human HTC75 cells induced by the truncated TRF2ΔBΔM protein are identified by white arrows. At the point of joining, only one chromatid from each of the two participating chromosomes enters into the fusion. All TCCs involve only leading-strand telomeres (identified by the G-rich telomere probe). (D) The top chromosome identified by yellow arrowheads has two leading-strand telomeres on the same chromatid. This “cis” configuration is the result of one (or any odd number) sister chromatid exchange, which are common events particularly in subtelomeric regions (23). The lower chromosome with yellow arrowheads displays the “trans” configuration. (E and F) Analysis of a mouse scid (DNA-PKcs–deficient) cell containing a telomeric chromatid fusion. (E) The G-rich probe identifies it as a leading-leading strand fusion. (F) The same cell after removal of the G-rich probe by denaturation and hybridization of the C-rich probe. A reciprocal pattern, as indicated by arrowheads at the point of fusion and at the ends of two chromosomes, is obtained and illustrates the exquisite strand specificity of the technique. The light blue fluorescence in centromeric regions is due to “lateral asymmetry” [see (24)]. Scale bars, 5 μm.

Table 1

Number of telomeric chromatid concatenates (TCCs) in HTC75 cells.

View this table:

TCCs also occur in cells with reduced DNA-PK activity, although they are far less frequent and produce a much milder phenotype. Inspection of 850 DNA-PKcs–deficient mouse fibroblasts (13) identified only 14 such chromatid-type events (Table 2). CO-FISH analysis revealed that all 14 were of the leading-to-leading strand variety. Again, this proportion differs significantly from that expected of random end-to-end fusion (P < 0.001). In contrast, no telomeric fusions were seen in 800 wild-type repair-proficient mouse fibroblasts.

Table 2

TCCs in five DNA-PK–deficient mouse cell lines.

View this table:

Conceivably, end-capping failure may affect only leading-strand telomeres because they have an absolute requirement for TRF2 and DNA-PKcs to refashion their blunt ends after replication. Lagging-strand telomeres already have 3′ overhangs after replication, so postreplicative processing may not be essential (but could occur to some extent). Both telomeres may then be folded into t loops as their final configuration. This interpretation is appealing for DNA-PKcs because of its role in DNA repair, which requires end processing. However, TRF2 is more commonly associated with remodeling chromosome ends into t loops (9, 10). If TRF2's role is confined to t loop formation, it would be difficult to explain the lack of lagging-strand telomeric fusions in cells expressing TRF2ΔBΔM. Alternatively, there may be essential differences in capping leading- and lagging-strand telomeres beyond remodeling ends into 3′ overhangs. At present, neither interpretation can be excluded.

The indispensable role played by chromosomal termini in maintaining the stable inheritance of genetic information is underscored by the severity of the phenotype associated with dysfunctional telomeres. Curiously, telomere dysfunction has itself opened a window into understanding these essential structures.

  • * To whom correspondence should be addressed. E-mail: egoodwin{at}


View Abstract

Navigate This Article