Telomere maintenance activity is a hallmark of cancer. In some telomerase-negative tumors, telomeres become lengthened by alternative lengthening of telomeres (ALT), a recombination-mediated DNA replication process in which telomeres use other telomeric DNA as a copy template. Using chromosome orientation fluorescence in situ hybridization, we found that postreplicative exchange events involving a telomere and another TTAGGG-repeat tract occur at remarkably high frequencies in ALT cells (range 28–280/100 metaphases) and rarely or never in non-ALT cells, including cell lines with very long telomeres. Like the ALT phenotype itself, the telomeric exchanges were not suppressed when telomerase was activated in ALT cells. These exchanges are telomere specific because there was no correlation with sister chromatid exchange rates at interstitial locations, and they were not observed in non-ALT Bloom syndrome cells with very high sister chromatid exchange rates.
Most human tumors acquire indefinite replicative capacity through the reactivation of telomerase (1) . However, some cancers use alternative lengthening of telomeres (ALT; Ref. 2 ), and recent evidence indicates that this mechanism is common in various types of cancer including osteosarcoma (3) and glioblastoma multiforme, the most common type of primary brain tumor in adults (4) . ALT is a recombination-based DNA replication mechanism that results in lengthening of telomeres (5, 6, 7) . Given the evidence that the telomerase and ALT mechanisms coexist in some tumors (3 , 4 , 8) and the possibility that treatment of telomerase-positive tumors with effective telomerase inhibitors will result in activation of ALT (9) , therapeutic approaches aimed at inhibiting telomerase activity may need to be complemented with others that target ALT. However, the molecular events implicated in ALT remain unexplored, partly because of the lack of a specific biochemical assay. Here, using chromosome orientation fluorescence in situ hybridization (CO-FISH), we found evidence for a remarkably high level of telomeric exchange in ALT cells that was seen rarely or not at all in cells that do not use the ALT mechanism.
Materials and Methods
WI38 VA13/2RA human cells, here referred to as VA13 and GM847 cells are SV40-immortalized human fibroblast cell lines that use the ALT mechanism (8) . GM847 expresses the telomerase RNA subunit, hTERC, whereas VA13 does not (10) . The construction of their telomerase-positive derivatives was described previously (11 , 12) . The tumor cell lines are derived from prostate carcinoma (DU145 and LNCaP), breast adenocarcinoma (SKBR-3), ovarian adenocarcinoma (SKOV-3), osteosarcoma (R970–5, U-2 OS, and SAOS-2) and fibrosarcoma (HT1080). All, except U-2 OS and SAOS-2, are telomerase positive. HT1080+POT1 (variant 1/clone 12) are telomerase-positive HT1080 human fibrosarcoma cells that have lengthened telomeres (mean terminal restriction fragment length of 15 kb) after transfection with variant 1 hPOT1 cDNA (13) . JFCF-6/2H+hTERT cells are an SV40-immortalized telomerase-positive human fibroblast line in which the terminal restriction fragments have been lengthened to an average of 20 kb after transfection with a human telomerase reverse transcriptase (hTERT) expression plasmid. 5 BJ, a foreskin primary fibroblast cell strain, was obtained at early passage from J. Smith (Baylor College of Medicine, Houston, TX) and was transfected with hTERT. 6 WI38+hTERT, a lung fibroblast cell line, was obtained from J. Campisi (Lawrence Berkeley National Laboratory, Berkeley, CA). NIH3T3 cells, 293T cells, three lymphoblastoid cell lines, and peripheral blood lymphocytes (from healthy donors) were obtained from the Centre d’Etude du Polymorphisme Humain collection. GM16859 is a primary fibroblast cell strain from a Bloom syndrome patient (obtained from Coriell Cell Repositories).
Culture Conditions and Metaphase Preparations.
Most cell lines were grown in either α-MEM (VA13, R970–5, U-2 OS, and SAOS-2) or DMEM (293T, NIH3T3, HT1080, HT1080+POT1, JFCF-6/2H+hTERT, BJ, BJ+hTERT, WI38+hTERT, GM847, DU145, U-2 OS, SKBR-3, 293T, and GM16859) supplemented with 10 or 15% fetal bovine serum. LNCaP, the lymphoblastoid cell lines and PHA-stimulated peripheral blood lymphocytes were cultivated in RPMI +10% fetal bovine serum. For CO-FISH and sister chromatid exchange (SCE) analyses, cells were incubated with BrdUrd (30 μm) for either one or two doubling times before 1- to 2-h incubation with colcemid (0.1 μg/ml) followed by hypotonic shock and fixation (methanol/acetic acid). Some cell cultures were treated with mitomycin C (10−6 M) for 1 h before the addition of BrdUrd (14) . Metaphase spreads were obtained by dropping suspensions of fixed cells onto clean glass slides. Preparations were used the next day for telomeric FISH.
CO-FISH and SCE Analysis
Slides were treated before hybridization as described by Cornforth and Eberle (15) . For telomeric FISH and analysis of fluorescent signals, the procedure we described previously was used (12) . In preparations from cells incubated for two doubling times, the concentration of 4′,6-diamidino-2-phenylindole was increased to 4 μg/ml to distinguish light and dark sister chromatids.
Results and Discussion
CO-FISH is a strand-specific hybridization technique commonly used to deduce the orientation of sequences along the chromosomes (16) . It is based on the specific degradation, in metaphase chromosomes, of the newly synthesized strand that has incorporated a base substitute (usually BrdUrd). When applied to telomeres, which have a characteristic strand asymmetry, it is expected that CO-FISH will show one signal per chromosome extremity (Fig. 1, A–B) ⇓ . What we observed in ALT cells, however, was that a substantial proportion of metaphases had many chromosome extremities bearing “double signals” (Fig. 1, C–D ⇓ ; Table 1 ⇓ ). The most likely explanation is that an exchange had occurred between a telomere of the chromosomes with double signals and other TTAGGG-repeat sequence subsequent to DNA replication. However, it was not possible to identify the potential donors of the second signal (for example, by finding extremities that had lost their telomeric signal) in the same metaphases, mainly because in ALT cells, chromosome extremities with no detectable telomere signals are common. To circumvent this obstacle, we examined the telomerase-positive derivatives of the GM847 and VA13 cell lines that were generated by transfecting them with expression constructs for TERT (GM847; Ref. 11 ) or TERT and TERC (VA13; Ref. 12 ). In these cells, ALT is still active but all extremities have telomeric signals detectable by FISH, indicating that telomerase has lengthened the shortest telomeres (11 , 12) . Using CO-FISH, we observed telomeric exchanges in an even higher proportion of metaphases than in the parental cell lines (Table 1) ⇓ . However, telomeric signals were detected on all chromosome extremities (not shown), indicating that the double signals do not result from the loss of signals from other chromosomes.
Double signals may be the result of exchange events occurring within the telomere, which both transfer and leave behind enough unsubstituted T2AG3 repeats to be detected by CO-FISH. Whether these exchanges take place between sister chromatids or between different chromosomes is not known. An additional possibility is that the exchanges may even involve the interaction of a telomere with the extrachromosomal TTAGGG-repeat DNA that is present in ALT cells (reviewed in Ref. 17 ). The most likely explanation for the increased detection of telomeric exchanges in telomerase-positive ALT cells is that elongation of short telomeres by telomerase facilitates the visualization of telomeric exchanges by CO-FISH, suggesting that these exchanges also involve relatively short telomeres.
An alternative interpretation for the occurrence of the double signals in CO-FISH experiments could be the presence of repeat tracts in which the G-rich/C-rich asymmetry of telomeric strands has been disrupted. It is conceivable that the double-stranded extrachromosomal TTAGGG-repeat DNA fragments in ALT cells may be substrates for non-homologous end joining, resulting in telomeric repeat tracts being added to unprotected chromosome extremities. Given that such fragments could be ligated in any orientation, the resulting product would be a patchwork of interspersed G-rich and C-rich sequences, readily detectable by CO-FISH on both sister chromatids of particular chromosome extremities. Although several of the cell lines examined here are of clonal origin (Table 1) ⇓ the chromosomes that exhibited telomeric exchanges varied from metaphase to metaphase (not shown). In contrast, CO-FISH double signals were detected at constant interstitial locations in some aberrant chromosomes carried by VA13 cells (Fig. 1C) ⇓ , probably the mark of past nonhomologous end-joining events between two chromosome extremities still carrying telomeric repeats.
As a further and more direct test of whether the double signals detected by CO-FISH result from G-rich repeats being present on the same DNA strand at the opposite extremities of the chromosome caused by putative ligation events, or result from telomeric exchanges, we followed the segregation of CO-FISH telomeric signals after a second round of DNA replication, as above (Fig. 2) ⇓ . In the case of G-rich tracts being present on the same strand at opposite extremities (and in the absence of SCE), the dark sister chromatid would appear to be labeled on both extremities. In fact, CO-FISH signals on both extremities of dark sister chromatids were extremely rare compared with the number of CO-FISH signals linked to the lighter sister chromatid with no detectable SCE event (Fig. 2) ⇓ . Interestingly, we also observed occasional double signals in these experiments, indicating the occurrence of telomeric exchanges after the second round of replication (Fig. 2) ⇓ .
This remarkable level of telomeric exchange was specific for ALT cells, because double signals were observed rarely or not at all in normal cell strains and in telomerase-positive cell lines (Table 1) ⇓ . To test whether this phenomenon is confined to cell lines immortalized in vitro, we analyzed by CO-FISH eight cell lines derived from human tumors, six of which are telomerase positive (HT1080, DU145, SKBR-3, R970–5, SKOV-3, and LNCaP) and two of which use ALT for telomere maintenance (U-2 OS and SAOS-2). We observed frequent telomeric exchanges in metaphase preparations only from the two ALT cell lines (Table 1 ⇓ ; Fig. 1, E–F ⇓ ), supporting the ALT specificity of this observation. CO-FISH analysis may thus constitute an additional assay for the presence of ALT in tumor cell lines, although it may be difficult to carry out in fresh material from cancer patients because it requires metaphase preparations.
Because ALT cells bear very long telomeres, we considered the possibility that telomeric exchanges may be triggered by these long structures independently from the ALT process. Telomerase-positive human cell lines with telomeres of about 40 kb not being available to test this hypothesis, we used mouse embryonic fibroblasts (NIH3T3), the telomeres of which, as judged by quantitative FISH, were at least as long as in ALT cells (not shown; 18 ). CO-FISH analyses detected double signals at chromosome extremities in a few NIH3T3 metaphases (Table 1) ⇓ . We also examined human cells with mean telomere lengths over 10 kb (BJ+hTERT and WI38+hTERT not shown), 15 kb (HT1080+POT1; Ref. 13 ), and 20 kb (JFCF/2H+hTERT, not shown). Again, CO-FISH analyses revealed only occasional double signals in these cells (Table 1) ⇓ . This very low frequency suggests that although long telomeres may be a substrate for telomeric exchange, additional factors are required to trigger the high rate of exchange seen in ALT cells.
Because the telomeric exchanges may be occurring between sister chromatids in ALT cells, they could be the result of the same type of DNA-damage response that leads to interstitial SCEs. Nevertheless, they are not observed in normal cells or telomerase-positive tumor cell lines with high SCE rates even when these rates have been increased by several fold in response to a short exposure to mitomycin C (Table 1) ⇓ . Furthermore, telomeric exchanges were not observed in primary fibroblasts or lymphoblastoid cell lines derived from Bloom syndrome patients, which have SCEs rates 10- or 12-fold higher than normal cells (Table 1) ⇓ . Together, these findings suggest that in most cells, telomeres either are not accumulating DNA damage or are prevented from undergoing exchange events, possibly as a consequence of telomere cap function. In ALT cells, on the other hand, the telomeres may be subject to a significant level of DNA damage (or some other alteration that results in “uncapping” and, therefore, in recognition of the telomere end as a site of DNA damage) and consequently have a strong SCE response. It is important to note that the SCE frequency is increased only at the telomeres of ALT cells, but not at interstitial locations (Table 1) ⇓ , including sites containing telomeric repeats (not shown). This observation is in agreement with a recent report showing that ALT cells do not have increased levels of homologous recombination within chromosomes (19) .
One hypothesis to explain the high frequency of telomeric exchanges in ALT cells would be that several components of the mitotic homologous recombination apparatus have been recruited to the telomere for ALT-mediated telomere lengthening. Presumably, one key outcome of this recruitment is that the proteins required for SCE are present at ALT telomeres. If the telomeric exchanges in ALT cells involve sister chromatids, they may be a side effect of a high concentration of SCE proteins at the telomere, rather than being a part of the ALT lengthening. On the other hand, the data do not exclude the possibility that the telomeric exchanges are interchromosomal or involve extrachromosomal TTAGGG-repeat fragments, in which case they may involve additional mechanisms specific to ALT. Additional experiments are needed to explore the relationship between telomere exchange events detected by CO-FISH and the telomere-lengthening mechanisms operating in ALT cells.
Grant support: J. A. Londoño-Vallejo was supported by Grant ARC-4779 and R. R. Reddel was supported by a Carcinogenesis Fellowship from the Cancer Council of New South Wales.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: J. Arturo Londoño-Vallejo, INSERM U434, 27 rue Juliette Dodu, 75010 Paris. Phone: 33-1-53725122; Fax: 33-1-53725128; E-mail:or Roger Reddel, Cancer Research Unit, Children’s Medical Research Institute, Westmead, New South Wales 2145, Australia. Phone: 61-2-96872800, Fax: 61-2-96872120; E-mail:
↵5 L. Colgin, P. Bonnefin, R. Reddel, unpublished observations.
↵6 S. Bacchetti, unpublished observations.
- Received December 24, 2003.
- Revision received February 7, 2004.
- Accepted February 12, 2004.
- ©2004 American Association for Cancer Research.