• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Mol Cell. Author manuscript; available in PMC Nov 19, 2007.
Published in final edited form as:
PMCID: PMC2081152

Human Cancer Cells Harbor T-Stumps, a Distinct Class of Extremely Short Telomeres


Using a modified single telomere length analysis protocol (STELA) to clone and examine the sequence composition of individual human XpYp telomeres, we discovered a distinct class of extremely short telomeres in human cancer cells with active telomerase. We name them “t-stumps”, to distinguish them from the well-regulated longer bulk telomeres. T-stumps contained arrangements of telomeric repeat variants and a minimal run of seven canonical telomeric TTAGGG repeats, but all could bind at least one TRF1 or TRF2 in vitro. The abundance of t-stumps was unaffected by ATM alteration, but could be changed by manipulating telomerase catalytic subunit (hTERT) levels in cancer cells. We propose that in the setting of active telomerase and compromised checkpoints characteristic of human cancer cells, t-stumps define the minimal telomeric unit that can still be protected by a TRF1/TRF2-capping complex and, further, that hTERT (or telomerase) may have a role in protecting t-stumps.


Telomeres consist of tandemly repeated DNA sequences bound by telomeric proteins. They not only facilitate DNA end replication, but also distinguish the ends of linear eukaryotic chromosomes from double-stranded DNA breaks and protect chromosome ends from fusion and recombination (Blackburn, 2001). In vertebrate cells, one telomeric DNA strand contains simple repeats of 5′-TTAGGG-3′ and is oriented 5′ to 3′ toward the chromosome end; the other strand contains 5′-TAACCC-3′ repeats and runs 5′ to 3′ toward the centromere. The G-rich telomere strand ends in a 3′ single-strand overhang that can form a t-loop structure to sequester the DNA terminus (Griffith et al., 1999).

In most organisms, telomeres are replenished by telomerase, a ribonucleoprotein complex that is minimally composed of a reverse transcriptase (hTERT) and an RNA subunit (hTER) that provides the short template sequence for the telomeric DNA repeats. Telomerase levels are high in human germ-line and stem cells, but often low or barely detectable in human somatic cells. Telomere dysfunction and telomerase activation has been implicated in human cancer progression (Blasco and Hahn, 2003). 80 – 90% of human tumors or tumor-derived cell lines have robust telomerase activity (Shay and Bacchetti, 1997). In human primary fibroblasts, the inability to replenish telomeres is proposed to cause replicative senescence, which can be bypassed by overexpressing telomerase (Bodnar et al., 1998). Conversely, overexpressing a dominant-negative hTERT causes gradual telomere shortening and eventual apoptosis in human cancer cells, and slightly accelerates senescence in human primary fibroblasts (Hahn et al., 1999) (Kim et al., 2003).

Several lines of evidence also suggest a protective function for telomerase (on or off telomeres) separable from its role in bulk telomeric DNA elongation. First, a hypomorphic hTERT could not extend bulk telomere length but significantly prolonged human primary cell life span, uncoupling a bulk telomere lengthening effect from a cell growth effect of telomerase (Zhu et al., 1999) (Kim et al., 2003). Second, knockdown of hTER in human cancer cells caused rapid growth inhibition before any detectable bulk telomere shortening (Li et al., 2004). Third, further knockdown of the low endogenous hTERT in primary human fibroblasts attenuated their DNA damage response (Masutomi et al., 2005). Finally, mouse TERT overexpression in a transgenic mouse model induced proliferation of hair follicle stem cells even in TER-deleted animals (Sarin et al., 2005).

In cells with robust telomerase activity, bulk telomere lengths are maintained within a well-defined length range, regulated by telomerase, and by telomere-associated proteins that regulate telomerase recruitment, telomere accessibility, DNA end replication and exonucleolytic degradation of telomeres. However, if yeast are genetically depleted for either telomerase or Tel1 (yeast ATM kinase homolog), or both, occasional greatly shortened telomeres that are functionally uncapped, and thus undergo fusions with an induced double-stranded DNA break, rapidly become detectable well before bulk telomere shortening (Chan and Blackburn, 2003).

Here, we report that a distinct class of extremely short telomeres, which we name “t-stumps”, exists in human cancer and transformed cells. We measured telomere length distributions by single telomere length analysis (STELA) (Baird et al., 2003). STELA allows the measurement of individual telomeres, and importantly, does not bias against the detection of short telomeres, unlike conventional bulk telomere length determination by telomere restriction fragment (TRF) analysis. While TRF analysis involves restriction digestion of genomic DNA followed by Southern hybridization to a probe containing the telomeric repeats, STELA measures chromosome-specific telomere length of individual telomeric DNA molecules sampled from within the telomere population, such as X and Y chromosome telomeres (XpYp telomeres) (Baird et al., 2003). Theoretically, every chromosome’s telomeres can be examined by STELA if appropriate subtelomeric primers can be found (Britt-Compton et al., 2006).

Using the original and modified STELA protocols, we discovered that t-stumps are conspicuous in telomerase-positive human cancer cells, as well as in transformed fibroblast cells lacking the p53 and pRb checkpoint pathways. The abundance of t-stumps could be altered by manipulating hTERT levels. In fibroblasts with intact checkpoints, with or without ectopically overexpressed telomerase, t-stumps were much rarer. Depletion of ATM alone had no effect on t-stump abundance. Notably, even the shortest t-stumps were capable of binding TRF2 and TRF1 in vitro. We propose that t-stumps define the shortest telomeres that can still be protected by a minimal functional unit of the telomeric capping protein complex and be tolerated in cells lacking checkpoint pathways, and that hTERT (or telomerase) may have a role in protecting t-stumps.


A distinct population of extremely short STELA PCR products in some telomerase-positive human cancer cell lines

We examined telomere length distribution in human cell lines using STELA. We tested both the published telorette linker (Baird et al., 2003) and a modified telorette linker which incorporates a restriction enzyme site (Figure 1A), thus allowing cloning of individual telomeres. In the STELA procedure, first a “telorette” linker, comprising a 20-nucleotide unique sequence followed by seven bases of telomeric repeat homology, is annealed to the G-rich strand overhang of telomeres. The telorette linker can then be ligated to the 5′ end of the C-rich strand of the duplex telomeric DNA if the telorette is annealed immediately adjacent to the end of the C-rich strand. After ligation, the genomic DNA is diluted down to a small population of amplifiable telomeric ends per reaction. Multiple aliquots are independently analyzed by PCR, using a “teltail” primer of same sequence as the unique region of the telorette linker, together with a primer specific for a given chromosome-specific subtelomeric sequence (for example, XpYp, a sequence in the common subtelomeric sequence on the short arm of the X and the Y chromosomes). Southern hybridization to a probe containing the subtelomeric sequence of the specified chromosome is performed to visualize the PCR products, which each represent the double-stranded region of the XpYp-specific telomere molecule (also containing a defined region of the subtelomeric DNA ~400 bases).

Figure 1
A distinct class of extremely short STELA products in cancer cells with robust telomerase activity. (A) Schematic representation of STELA (Baird et al., 2003). Shaded small rectangles in the modified telorette linker and teltail primer represent the EagI ...

We compared the modified and original telorette linkers in parallel reactions using DNA of human primary fibroblasts (IMR90 line) and several representative telomerase-positive human cancer cell lines (Figure 1B). Since it had been reported that the distal ends of the human C-rich telomere strands preferentially terminate with the sequence 5′-CTAACC-3′ (Sfeir et al., 2005), in the majority of our assays we used telorette3, whose telomere-complementary sequence allows it to ligate specifically to telomeres with such ends. Both paired sets of telorette linkers allowed efficient amplification of bulk telomeres in IMR90 cells (Figure 1C, left panel), with a size distribution that agreed well with that reported by Baird et al for such senescing late-passage cultured fibroblasts (Baird et al., 2003).

STELA product distributions of the bulk telomeres from HeLa cervical cancer, T24 bladder cancer, H1299 lung cancer, HT1080 fibrosarcoma and UMUC3 bladder cancer cell lines, lines with highly active telomerase (Figure 1B), correlated well with those from the conventional TRF analysis (Figure 1C and D). Interestingly, in some lines extremely short STELA PCR products also appeared (Figure 1C), with a tight size distribution clustered around ~500bp. Since these telomeric STELA PCR products include the ~400bp subtelomeric region, this size suggested class of telomeres with telomeric repeat tract lengths minimally of ~100bp, which was confirmed in the analyses described below. In STELA analyses, individual bands represent single telomeric DNA molecules sampled in the diluted genomic DNA samples used for the assay reactions; hence the relative abundances of products can be compared among samples. The abundance of these extremely short STELA products was highly cell line-specific: low in IMR90 (including at senescence), HeLa and T24 cells, and higher in H1299, HT1080 and UMUC3 cells, where their abundance was comparable to that of telomeres in the lower end of the bulk long telomere population range.

Several lines of evidence ruled out the possibility that these very short STELA products were a PCR-based or other artifact. First, since in the STELA procedure genomic DNA was digested with EcoRI to facilitate solubilization before the telorette annealing and ligation step, it was conceivable that the short products had been generated by variant telomeric sequences that contained an EcoRI site and hence were cut by this restriction enzyme. However, short STELA products were detected when the genomic DNA of UMUC3 cells was digested with a different restriction enzyme, XhoI, or not digested (Figure 2A and 2B). Second, the short STELA products were still detected using the telorette linker telorette4-Mod1 that ends with a different base in the telomeric homology region, or the telorette linker telorette3-Mod2 that has a different unique sequence region (Figure 2C). This eliminated the possibility of primer-specific artifactual products. Third, a chromosome 9-specific subtelomeric primer produced similar short STELA products from genomic DNA of UMUC3 cells with or without telomere elongation via hTER overexpression (Figure 2D and Figure 6B).

Figure 2
Validation of the extremely short telomeric STELA products using UMUC3 cells. (A) STELA was performed on genomic DNA pre-treated with XhoI. (B) STELA was performed on genomic DNA that was not pre-treated with any restriction enzyme. (C) STELA was performed ...
Figure 6
The abundance of t-stumps in cancer cells can be changed by altering relative levels of telomerase components. (A) STELA of human IMR90 primary fibroblasts. IMR90 cells were infected with pBabe-puro or pBabe-puro-hTERT retroviruses at PD43. Puromycin-resistant ...

Validation of the extremely short STELA products as telomeres by telomere cloning

To verify that the extremely short STELA products were truly of telomeric origin, and to examine the exact sequences of XpYp telomeres, we modified the STELA procedure in order to clone and sequence the STELA products: we carried out STELA PCR with genomic DNA from UMUC3 cells using telorette3-Mod1 that contained a restriction enzyme site at the tail region. The resulting PCR products were digested with EagI and PstI, which cut the tail of the telorette linker and the XpYp-specific subtelomeric region respectively (see Figure 1A), ligated into a cloning vector, and used to transform bacteria cells. We then performed colony hybridization using an XpYp-specific subtelomeric probe to identify colonies transformed with plasmids containing XpYp-specific subtelomeric sequences. DNA was extracted from the colonies and a total of 30 cloned STELA products analyzed and sequenced. All 30 clones contained the XpYp subtelomeric segment followed by telomeric repeats, validating the telomeric identity of the STELA products. The length distribution of the telomere repeat tracts was bimodal (Figure 3A and Supplemental Figure 1): 9 clones contained extremely short telomeric repeat tracts of just 90–300 bases abutting the subtelomeric sequence and 10 clones had telomeric repeats of over 1300 bases, while the 11 remaining clones had telomeric repeat tracts ranging from 300bp to 1300bp. The plasmids with telomere repeat lengths longer than 1300 bases tended to recombine in E. coli, consistent with previous reports that human and other telomeric sequences are unstable in bacteria cells (de Lange et al., 1990; Henderson and Blackburn, 1989). Taken together, these telomere cloning results validated the STELA approach and confirmed that the extremely short products revealed by STELA analysis consist of telomeric sequences, as described further below. We named these extremely short telomeres “t-stumps” to distinguish them from the longer bulk telomeres.

Figure 3
Sequence composition of cloned XpYp telomeres of UMUC3 cells. (A) Histogram of cloned telomeres grouped by telomere tract length (STELA product length minus ~400bp subtelomeric sequence). (B) Sequences of nine extremely short telomeres cloned. Sequence ...

Sequence arrangements of inner (centromere-proximal) telomeric repeats and t-stumps

Sequence analysis revealed several features of telomeric repeat tracts (Figure 3B and C). First, all 30 cloned XpYp telomeres, long or short, contained a common sequence block consisting of a run of variant telomeric repeats 5′-(TCACCC)6TCACCTTCACCC-3′ (to be referred to as the “TCACCC segment”), immediately distal to and adjacent to the unique subtelomeric sequence. Second, the shortest t-stump cloned contained only 7 tandem uninterrupted canonical telomeric 5′-TAACCC-3′ repeats immediately adjacent to the TCACCC segment. Third, 4 of the 9 t-stump clones (Figure 3B) had a variant sequence, 5′-TAACCCCCC-3′, in the middle of the canonical TAACCC repeats, residing at the same place distal to the subtelomeric region. Fourth, some longer t-stump clones had other variant repeats containing G residues toward the distal telomeric end (Figure 3B). Finally, the longer bulk telomere-length clones contained an additional long tract of canonical, uninterrupted TAACCC repeats toward the distal telomeric end (Figure 3C). Because of the difficulty in sequencing through repetitive sequences and G-rich sequences, we cannot exclude the possibility that TAACCCCCC variant repeat or G-residue-containing sequences also occur at the more centromere-proximal region of such long telomere clones as well.

Hence, human telomerase–positive cancer cells, in addition to the bulk telomeres, contain a distinct population of extremely short telomeres, the “t-stumps”, which are relatively enriched for variant telomeric repeats compared to long bulk telomeres, most of whose outer tracts consist of the canonical perfect TAACCC repeats.

The shortest t-stumps can bind telomere-protective proteins

The sequence-specific, TTAGGG repeat-binding factors TRF1 and TRF2 bind duplex telomeric DNA repeat tracts in vertebrate cells (de Lange, 2005). Both are involved in telomere length maintenance. TRF2 additionally protects chromosome ends from end-to-end fusions. Its loss results in chromosome abnormalities and activation of ATM- and p53-dependent DNA damage signaling (Karlseder et al., 1999). To determine the potential of t-stumps for chromosome end protection, we carried out gel-shift analysis with several representative cloned t-stumps, using in vitro-translated TRF1 and TRF2 proteins. First, as expected, the XpYp-subtelomeric region alone showed no binding to TRF1 or TRF2. The common “TCACCC segment”, located between the subtelomeric region and the distal canonical TAACCC telomeric repeats, when subcloned out also failed to bind either TRF1 or TRF2 in vitro (Figure 4A and B). However, addition of the 7 abutting TAACCC repeats - which together with the TCACCC segment comprise the shortest cloned t-stump - was sufficient to confer efficient binding of at least one molecule of TRF1 or TRF2 (Figure 4A and B). Longer cloned t-stumps in the gel-shift assays, despite their having G residues in the middle of the terminal TAACCC repeats, showed additional shifted bands, presumably representing t-stump DNAs bound by increasing numbers of TRF1 or TRF2. A recent report demonstrated that a telomeric complex containing only 12 TTAGGG repeats and TRF2/Rap1 could inhibit non-homologous end-joining (NHEJ) using an in vitro NHEJ assay (Bae and Baumann, 2007), consistent with our results. Thus, t-stumps may represent the shortest telomeres that can still be bound by TRF1 and TRF2, and by inference be protected by TRF2 capping from eliciting a DNA damage response.

Figure 4
T-stumps bind TRF1 and TRF2 in vitro. (A) Gel-shift assays with in vitro-translated TRF2 and 32P-labeled t-stump clones. (B) Gel-shift assays with in vitro-translated TRF1 and 32P-labeled t-stump clones. The subtelomeric DNA used in the assays is a DNA ...

Apoptosis does not alter the abundance of t-stumps

Having established the authentic telomeric origin of t-stumps, we then investigated their potential source and factors that affect their abundance. To test whether t-stumps result from breakdown of telomeres in apoptotic cells in the cancer cell population, apoptosis was induced in two different ways: Overexpressing wildtype p53, as expected (Gomez-Manzano et al., 1996) caused high apoptosis levels (Figure 5A and B) in UMUC3 cells, which contain a mutated p53 (Mizuarai et al., 2006). However, the abundance of t-stumps remained similar. We also expressed mutant-template telomerase RNA to induce apoptosis by uncapping telomeres, as previously shown in multiple cancer cell lines (Supplemental Figure 2A, B) (Li et al., 2004) (Xu and Blackburn, 2004). UMUC3 cells overexpressing two different mutant-template telomerase RNAs each had significantly higher apoptotic populations than controls, but did not have more t-stumps (Supplemental Figure 2A and C). Taken together, these results rule out the possibility that apoptosis is the source of t-stumps.

Figure 5
Induction of apoptosis does not affect the abundance of t-stumps. UMUC3 cells were infected with recombinant adenoviruses expressing GFP or wildtyple p53 at 40 pfu/cell. Eighteen hours after infection, cells were collected for STELA analysis. (A) Phase ...

The abundance of t-stumps in cancer cells can be manipulated by altering levels of hTERT

The TERT of yeast (called Est2) associates with yeast telomeres in vivo even in G1, a time in the cell cycle when telomerase is incapable of enzymatic elongation of telomeric DNA (which is confined to S and G2/M phases of the cell cycle in that species) (Smith et al., 2003; Taggart et al., 2002). In addition, in budding yeast Candida albicans, telomerase protects telomeres from degradation even in the absence of telomere elongation (Hsu et al., 2007). We first tested the effect of telomerase manipulation on t-stumps by overexpressing hTERT in senescing human primary IMR90 fibroblast cells, which normally have extremely low telomerase activity. Consistent with previous reports (Bodnar et al., 1998) (Kim et al., 2003), overexpression of hTERT expanded the replicative life span of IMR90 cells and significantly extended bulk telomere length (data not shown). We had found that in senescing IMR90 cells at population doubling 70 (PD70) (Kim et al., 2003), t-stumps were present, although at relatively low levels, and with a wider (longer) size distribution than in transformed or cancer cells (Figure 1C and and6A).6A). This agrees with the STELA telomere length distribution reported for another senescent primary fibroblast line (Baird et al., 2003). The size distribution of the short telomeres was not changed by hTERT overexpression (Figure 6A). It was proposed that in normal primary cells the rare short telomeres (whose tract lengths are estimated to be less than 100bp) trigger a DNA damage response and the onset of senescence (Baird et al., 2003). Therefore, it was striking that the abundance and size distributions of t-stumps in the robustly growing hTERT-overexpressing fibroblasts at PD99 were very similar to those in senescent cells (Figure 6A). Hence, when this primary cell line overexpresses hTERT, similarly extremely short telomeres do not elicit senescence signaling, even though all cell checkpoint and damage response pathways are expected to be intact. Such a finding could be explained by a protective function for hTERT at telomeric ends, as discussed below.

To examine the effect of manipulating telomerase components on the abundance of t-stumps in cancer cells, we first overexpressed either the telomerase RNA subunit hTER, or the catalytic subunit hTERT, in UMUC3 cells. Unusually, UMUC3 cells have their telomerase activity limited by hTER level, rather than the more common situation of limitation via hTERT level: overexpression of hTER in UMUC3 cells significantly lengthened the bulk telomeres from 1.6–5 kb to 3–10 kb and increased telomerase activity, measured by in vitro assays of the cell extracts, by 13-fold; in contrast, overexpression of hTERT did not detectably lengthen bulk telomeres, and increased in vitro telomerase activity only about 2-fold (Figure 6B and Supplemental Figure 3B). Therefore UMUC3 cells are an advantageous system for dissecting the role of hTERT in t-stump formation or stabilization from its role in telomere elongation in cancer cells. As shown in Figure 6B and 6C, the bulk telomere length distribution seen by STELA analysis faithfully recapitulated the conventional telomere restriction fragment length analysis: overexpression of hTER increased the bulk telomere length whereas overexpression of hTERT did not. Interestingly, in these UMUC3 cells, overexpression of hTERT increased the abundance of t-stumps (Figure 6C).

In contrast to UMUC3 cells, overexpression of hTER in HT1080 cells caused no detectable effect on bulk telomere length (Figure 6B and 6D). However, overexpression of hTERT led to moderate lengthening of bulk telomeres from 2–11kb to 2.5–12kb, suggesting that hTERT is limiting telomerase activity. In the HT1080 cells overexpressing hTERT, the lower size range of the bulk telomeres shifted upwards and t-stumps became significantly less abundant (Figure 6D). It has been demonstrated in budding yeast and mouse that the shortest telomeres are elongated preferentially by telomerase (Zhu et al., 1998) (Hemann et al., 2001) (Samper et al., 2001) (Teixeira et al., 2004). Hence the increased active telomerase in the hTERT-overexpressing HT1080 cells might potentially elongate the t-stumps and/or protect them from eliciting a catastrophic DNA damage response.

We confirmed the effects of hTERT on t-stumps by fluorescence in situ hybridization with a peptide-nucleic acid (PNA)-telomere probe (telomere-FISH), performed on UMUC3 parental control cells or UMUC3 cells overexpressing hTER or hTERT. The extremely short t-stumps should show up as signal-free ends because of the lower detection limit of the telomeric-FISH technique. We scored signal-free ends from 30 metaphase spreads for each line. Indeed overexpression of hTERT resulted in higher levels of signal-free ends per metaphase than parental control cells (MeanhTERT=2.83, Meancontrol=1.47, P<0.0067) (Figure 6E and F). In contrast, although overexpression of hTER lengthened the bulk telomeres (Figure 6B), there was no significant difference in the frequency of signal-free ends per metaphase between these cells and the parental control cells (MeanhTER=1.57, Meancontrol=1.47, P<0.8).

Effects of altering checkpoint pathways on t-stump abundance

Shortened telomeres have been reported to trigger ATM/ATR-dependent DNA damage responses and the onset of senescence in human primary fibroblasts (d’Adda di Fagagna et al., 2003) (Herbig et al., 2004). Dysfunctional telomeres caused by overexpression of a dominant-negative allele of TRF2 trigger an ATM- and p53-dependent apoptotic response (Karlseder et al., 1999). In addition, lack of Tel1 (ATM homolog) or telomerase in budding yeast cells leads to an increase in the abundance of greatly shortened telomeres that by definition had become uncapped, as they were detected because they had fused to a DNA double-strand break (Chan and Blackburn, 2003). DNA damage checkpoint pathways are intact in normal human cells but not in human cancer cells. Interestingly however, abrogating the ATM-dependent DNA damage checkpoint did not alter t-stump accumulation, as shown in three different experimental settings. First, both the size distribution and very low abundance of t-stumps were similar in normal and ATM-deficient human primary fibroblast cells at similar population doublings (Figure 7A). Second, in both an ATM-wildtype and an ATM-deficient SV40-transformed cell line, a large accumulation of t-stumps was apparent, which again was similar in both the lines shown in Figure 7B, and in two additional independent ATM wildtype and ATM-deficient fibroblast lines transformed by SV40 (data not shown). Finally, ablation of ATM expression by RNA interference via two different ATM-targeting shRNAs did not change t-stump abundance in UMUC3 cells (Figure 7C). In conclusion, impaired ATM-function is not sufficient to cause t-stump accumulation in primary fibroblasts, and also had no effect on the t-stump levels in SV40-transformed or cancer cells.

Figure 7
Lack of p53 and pRb function, but not of ATM function, affects t-stump accumulation. (A) STELA of ATM-wildtype IMR90 and ATM-deficient GM02052C primary fibroblast cells. (B) STELA of SV40 immortalized lines GM00637J (ATM-wildtype) and GM09607B (ATM-deficient). ...

SV40-transformation inactivates the p53 and pRb checkpoint pathways. In all SV40-transformed fibroblast cell lines examined, t-stumps were more abundant than in non-transformed primary fibroblasts (compare Figure 7B with 7A). In these SV40-transformed cells, telomerase was also active (Supplemental Figure 3C). It was not possible to perform STELA to examine the effect of telomerase expression on t-stump abundance in SV40-transformed telomerase-negative (ALT) cells. This was because the very long bulk telomeres (>20kb) precluded PCR with reproducible efficiency, and hence any analysis of relative amounts of bulk versus t-stump size classes.

Taken together, these data suggest that impairing the functions of p53 and pRb in cells with active telomerase allows t-stumps to accumulate. In contrast, loss of ATM-function alone is not sufficient to allow t-stump accumulation.


Here we have reported the discovery of a distinct class of extremely short telomeres, which we name “t-stumps”, in transformed human cells containing active telomerase. In contrast, similarly short telomeres reported previously in senescing primary human cells and human germ line were rare (Baird et al., 2006). The cancer cell t-stumps, while enriched for variant telomeric repeats, may define a telomeric sequence that can bind a minimal telomeric protective protein unit. The ability of cells to tolerate a certain level of t-stumps can be changed by manipulating levels of telomerase components. Our finding suggest that t-stumps are tolerated in checkpoint-defective cells and that at least one component of telomerase, hTERT, might have a protective function at t-stumps separable from its function in elongating telomeric DNA.

Telomeric repeat variants in human chromosomes

Southern hybridization of human telomeres against different oligonucleotides indicated that variant telomeric DNA repeats cluster within the centromere-proximal portion of human telomeric tracts (Allshire et al., 1989). Telomere variant repeat mapping PCR (TVR-PCR), designed based on the predicted variant repeats, was used to map the distribution pattern of these variant repeats (Baird et al., 1995). However, such hybridization- or PCR-based approaches cannot reveal the exact sequence composition of telomeres. To further characterize the sequence arrangement of human telomeres, we modified the STELA procedure for cloning a large number of human XpYp telomeres directly into a bacterial plasmid vector. Human telomeres had previously been cloned by various approaches. One approach (Brown, 1989) involved enriching human telomeric DNA by isopycnic ultracentrifugation on Ag+/Cs2SO4 gradients, followed by ligation of telomere fragments with a yeast minichromosome and recovery from yeast cells. Another direct cloning approach (de Lange et al., 1990) involved treating genomic DNA with a combination of restriction enzymes, followed by isolating DNA fragments larger than 10kb (which is enriched for telomeric fragments since telomeres generally lack restriction sites). The DNA fragments were then blunt-ended, ligated to linkers and cloned into a bacterial vector. Positive colonies were screened using a telomeric probe. Compared to the aforementioned telomere cloning strategies, our STELA analysis-based human telomere cloning method takes less preparation, is easier to carry out and yields more positive clones per screen. It should be noted that the STELA method results in analysis only of the duplex portion of the telomeric tracts, and while it relies on the G-strand overhang it does not address the properties of that single-stranded end portion of the telomere. Nonetheless, a crucial advantage of this telomere cloning method is that the integrity of the C-strand telomere ends is preserved, unlike the yeast minichromosome approach in which the ends are subject to modification in the yeast cells or the direct cloning approach in which the ends must be polished by an exonuclease for ligation of linkers. As more human genomic sequences are becoming available, it will be possible to define more chromosome-specific subtelomeric primers, clone out telomeres on different chromosomes and scrutinize their sequence arrangements.

Sequence analysis of XpYp-specific telomeres of UMUC3 cells revealed a common run of ~8 tandem 5′-TCACCC-3′ repeats (the “TCACCC segment”) between the subtelomeric region and the canonical telomeric sequence in all cloned telomeres. Such arrangements of variant telomeric sequences also occur in the inner regions of telomeres of other species. Examples include the perfect TTTGGGG repeat tracts instead of the canonical TTGGGG repeats in the inner portion of germline chromosomes of the protozoan Tetrahymena (Kirk and Blackburn, 1995) and comparably positioned variants of the canonical TTTAGGG repeats in the telomeres of the plant Arabidopsis thaliana (Richards et al., 1992). We speculate that such a variant repeat run may arise through replication slippage, telomerase misincorporation or recombination. The fact that this variant repeat tract exists at every cloned XpYp telomere in UMUC3 cells indicates that it is relatively stable and not subjected to frequent additional replication slippage or mutation during propagation of the cell line. The recurrence of common sequence blocks of other variant repeats in variable arrangements in t-stumps prompts the speculation that these arose by recombination in the inner regions of telomeric tracts. However, even the shortest t-stumps analyzed bound TRF1 and TRF2, suggesting that they define the minimal functional telomeric unit.

Checkpoint pathways and t-stumps

While t-stumps are rare in telomerase-positive germ line cells (Baird et al., 2006), as well as in hTERT-overexpressing primary fibroblasts and senescing telomerase-negative primary fibroblasts, we have shown they are present in relatively high amounts in telomerase-positive cancer cells and transformed cells. Expression of mutant-template telomerase RNA under conditions that cause uncapping of a high fraction of cancer cell telomeres had no effect on t-stump abundance (Supplemental Figure 2B, C), and it will be of interest to determine if depletion or overexpression of specific components of the telomeric protein complex affect t-stump abundance. Our results with ATM-deficient primary cells and ATM knockdown in cancer cells indicated that the DNA damage sensor protein ATM does not play a major role in allowing t-stumps to accumulate in a cell population. On the other hand, those cancer cell lines with relatively more abundant t-stumps have a variety of checkpoint defects. For example, they vary in their p53 status - wildtype (HT1080), null (H1299), or mutated (UMUC3) (Peng et al., 2001) (Mizuarai et al., 2006). Other components in the p53 pathway, as well as other checkpoint pathways such as the p16 pathway, are often deficient in human cancer cells. For example, the HT1080, H1299 and UMUC3 cells are all p16-deficient (Grim et al., 1997; Roninson, 2003). Thus the p53 and/or p16 pathways are potential candidates for involvement in determining the response of a cell to t-stumps. Indeed, SV40-transformed fibroblasts, in which p53 and pRb functions are abrogated, had greatly increased t-stump abundance. SV40-transformed primary human cells continue to divide past the usual senescence point while their telomeres keep eroding until the point of cell crisis, when significant cell death ensues. At crisis, the mean telomere length is much shorter than that at senescence (Counter et al., 1992), suggesting that loss of p53 and pRb checkpoints compromises the ability of cells to detect short telomeres and initiate a DNA damage response and the onset of replicative senescence. Our finding that t-stumps accumulated in immortalized telomerase-positive fibroblast cells in which the p53 and pRB pathways were blocked as a result of SV40 transformation is consistent with this hypothesis.

Telomerase and t-stumps

An interplay between telomerase components and t-stump abundance is suggested by the following evidence: First, telomerase-positive cancer cells often have more abundant t-stumps. Second, in HT1080 cells, in which hTERT level limits telomerase activity, t-stumps became less abundant upon hTERT overexpression. This suggests that when extra hTERT protein can make an active telomerase RNP complex together with the available RNA subunit hTER, and further lengthen telomeres, including t-stumps, the t-stump abundance is lowered. In contrast, in UMUC3 cells, in which hTERT is normally already in excess (because its overexpression does not increase bulk telomere length), hTERT overexpression elevated t-stump abundance. This result suggests that the excess hTERT can either protect and stabilize t-stumps, or actively promote their generation.

The STELA assay measures steady-state levels of t-stumps, and thus alterations of t-stump levels could reflect differences in relative or absolute rates of generation or loss of t-stumps. Three types of models could explain the effects of hTERT on the steady-state levels of t-stumps. In the first model, t-stumps are generated at a constant rate unaffected by hTERT, but hTERT (or telomerase), along with TRF2 (and/or TRF1), binds and stabilizes the t-stumps, thus protecting them from degradation. (These t-stumps may at some point also become substrates for elongation by telomerase). The current lack of an available chromatin immunoprecipitation (ChIP)-grade hTERT antibody limits our ability to test this model directly. In the second model, hTERT does not affect the rate of formation of t-stumps or the rate of their loss (via conversion to long telomeres or by degradation), but rather, protects cells containing t-stumps from being eliminated from the cell population. This may be achieved through hTERT binding to the t-stumps (as in the first model) or other effects on cell signaling pathways. A third formally possible model is that hTERT promotes the generation of t-stumps.

In S. cerevisiae, telomerase protects telomeres from non-homologous end-joining to an induced double-stranded break (Chan and Blackburn, 2003). In telomerase-defective cells, in which such fusion frequencies increased, the fusion junction contained very little telomeric DNA, akin to a t-stump and indicative of catastrophic telomere shortening preceding the fusion. In light of our findings in human cancer cells, we suggest that in the normally constitutively telomerase-positive yeast cells, extremely short telomeres might similarly exist but be protected by telomerase from undergoing fusions. However, when telomerase function is compromised, these extremely short telomeres become deprotected and thus undergo the observed fusions.

Knockdown of telomerase RNA (hTER) leads to a rapid growth inhibition of cancer cells that does not require p53 or bulk telomere shortening (Li et al., 2004). The existence of t-stumps in telomerase-positive cancer and transformed cells might be relevant to this result; the depletion of hTER (and the consequent reduction of telomerase) in these cells may have rapidly deprotected t-stumps, which resulted in them being perceived as dysfunctional telomeres and hence elicited the growth inhibition response.

In summary, we have found that t-stumps of a very short size class accumulate only in telomerase-containing cells (cancer or SV40-transformed cells) that also lack checkpoint pathways. Hence, we propose that a situation of simultaneously high telomerase and abrogated DNA damage checkpoint pathways (involving p53 and/or pRb) makes cells surprisingly tolerant of these minimal telomeric entities.

Experimental Procedures

See Supplemental Data for additional experimental information.

Cell Lines

IMR90 (primary fibroblast), HeLa (cervical carcinoma), T24 (bladder carcinoma), H1299 (lung carcinoma), HT1080 (fibrosarcoma) and UMUC3 (bladder carcinoma) were obtained from the ATCC. GM02052C (ATM-deficient primary fibroblast), GM00637J (ATM-wildtype transformed fibroblast) and GM09607B (ATM-deficient transformed fibroblast) were obtained from the Coriell Institute.

STELA Oligonucleotides

STELA was carried out essentially as described (Baird et al., 2003) (see Supplemental Data for details). Telorette3, 5′-TGCTCCGTGCATCTGGCATCCCTAACC-3′; telorette3-Mod1, 5′-TGCTCGGCCGATCTGGCATCCCTAACC-3′; telorette3-Mod2, 5′-TGCTCGAGGCATCTGGCATCCCTAACC-3′; telorette4-Mod1, 5′-TGCTCGGCCGATCTGGCATCCTAACCC-3′; teltail, 5′-TGCTCCGTGCATCTGGCATC-3′; teltail-Mod1, 5′-TGCTCGGCCGATCTGGCATC-3′; teltail-Mod2, 5′-TGCTCGAGGCATCTGGCATC-3′; 6176, 5′-GTAAGCACATGAGGAATGTGG; 6177, 5′-AAGGCGGAGCAGAGTTCTCG-3′; XpYpE2, 5′-GTTGTCTCAGGGTCCTAGTG-3′, our XpYpE2 primer has an addtional G at the 5′-end of the original XpYpE2 primer. The change was found to enhance overall STELA signal intensity significantly; XpYpB2, 5′-TCTGAAAGTGGACC(A/T)ATCAG-3′.

Telomere Cloning

STELA analysis was carried out using telorette3-Mod1 in the ligation reaction and the primer pair XpYpE2 and teltail-Mod1 in the PCR reaction. PCR products were extracted with phenol/chloroform. DNA was precipitated before digestion with EagI and PstI. After phenol/chloroform extraction, oligonucleotide primers were removed from the reaction mixture by a SizeSep spin column (Amersham). Eluted DNA was precipitated and ligated to an EagI- and PstI-digested, dephosphorylated pBluescript vector. Transformed DH5αcells were grown at 30°C for 24h. Colony hybridization against a probe generated by PCR from XpYpE2 and XpYpB2 was carried out to screen for clones containing XpYp-specific subtelomeric sequences.

Supplementary Material



We thank Dr. Bradley Stohr for lentiviral shRNA constructs against ATM. We thank Dr. Jiandong Chen for recombinant adenoviruses expressing p53 or GFP. We thank Dr. Bradley Stohr and Dr. Wallace Marshall for critical reading of the manuscript. This study was supported by NCI grant CA96840 and the Bernard Osher Foundation.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Allshire RC, Dempster M, Hastie ND. Human telomeres contain at least three types of G-rich repeat distributed non-randomly. Nucleic Acids Res. 1989;17:4611–4627. [PMC free article] [PubMed]
  • Bae NS, Baumann P. A RAP1/TRF2 complex inhibits nonhomologous end-joining at human telomeric DNA ends. Mol Cell. 2007;26:323–334. [PubMed]
  • Baird DM, Britt-Compton B, Rowson J, Amso NN, Gregory L, Kipling D. Telomere instability in the male germline. Hum Mol Genet. 2006;15:45–51. [PubMed]
  • Baird DM, Jeffreys AJ, Royle NJ. Mechanisms underlying telomere repeat turnover, revealed by hypervariable variant repeat distribution patterns in the human Xp/Yp telomere. Embo J. 1995;14:5433–5443. [PMC free article] [PubMed]
  • Baird DM, Rowson J, Wynford-Thomas D, Kipling D. Extensive allelic variation and ultrashort telomeres in senescent human cells. Nat Genet. 2003;33:203–207. [PubMed]
  • Blackburn EH. Switching and signaling at the telomere. Cell. 2001;106:661–673. [PubMed]
  • Blasco MA, Hahn WC. Evolving views of telomerase and cancer. Trends Cell Biol. 2003;13:289–294. [PubMed]
  • Bodnar AG, Ouellette M, Frolkis M, Holt SE, Chiu CP, Morin GB, Harley CB, Shay JW, Lichtsteiner S, Wright WE. Extension of life-span by introduction of telomerase into normal human cells. Science. 1998;279:349–352. [PubMed]
  • Britt-Compton B, Rowson J, Locke M, Mackenzie I, Kipling D, Baird DM. Structural stability and chromosome-specific telomere length is governed by cis-acting determinants in humans. Hum Mol Genet. 2006;15:725–733. [PubMed]
  • Brown WR. Molecular cloning of human telomeres in yeast. Nature. 1989;338:774–776. [PubMed]
  • Chan SW, Blackburn EH. Telomerase and ATM/Tel1p protect telomeres from nonhomologous end joining. Mol Cell. 2003;11:1379–1387. [PubMed]
  • Counter CM, Avilion AA, LeFeuvre CE, Stewart NG, Greider CW, Harley CB, Bacchetti S. Telomere shortening associated with chromosome instability is arrested in immortal cells which express telomerase activity. Embo J. 1992;11:1921–1929. [PMC free article] [PubMed]
  • d’Adda di Fagagna F, Reaper PM, Clay-Farrace L, Fiegler H, Carr P, Von Zglinicki T, Saretzki G, Carter NP, Jackson SP. A DNA damage checkpoint response in telomere-initiated senescence. Nature. 2003;426:194–198. [PubMed]
  • de Lange T. Shelterin: the protein complex that shapes and safeguards human telomeres. Genes Dev. 2005;19:2100–2110. [PubMed]
  • de Lange T, Shiue L, Myers RM, Cox DR, Naylor SL, Killery AM, Varmus HE. Structure and variability of human chromosome ends. Mol Cell Biol. 1990;10:518–527. [PMC free article] [PubMed]
  • Gomez-Manzano C, Fueyo J, Kyritsis AP, Steck PA, Roth JA, McDonnell TJ, Steck KD, Levin VA, Yung WK. Adenovirus-mediated transfer of the p53 gene produces rapid and generalized death of human glioma cells via apoptosis. Cancer Res. 1996;56:694–699. [PubMed]
  • Griffith JD, Comeau L, Rosenfield S, Stansel RM, Bianchi A, Moss H, de Lange T. Mammalian telomeres end in a large duplex loop. Cell. 1999;97:503–514. [PubMed]
  • Grim J, D’Amico A, Frizelle S, Zhou J, Kratzke RA, Curiel DT. Adenovirus-mediated delivery of p16 to p16-deficient human bladder cancer cells confers chemoresistance to cisplatin and paclitaxel. Clin Cancer Res. 1997;3:2415–2423. [PubMed]
  • Hahn WC, Stewart SA, Brooks MW, York SG, Eaton E, Kurachi A, Beijersbergen RL, Knoll JH, Meyerson M, Weinberg RA. Inhibition of telomerase limits the growth of human cancer cells. Nat Med. 1999;5:1164–1170. [PubMed]
  • Hemann MT, Strong MA, Hao LY, Greider CW. The shortest telomere, not average telomere length, is critical for cell viability and chromosome stability. Cell. 2001;107:67–77. [PubMed]
  • Henderson ER, Blackburn EH. An overhanging 3′ terminus is a conserved feature of telomeres. Mol Cell Biol. 1989;9:345–348. [PMC free article] [PubMed]
  • Herbig U, Jobling WA, Chen BP, Chen DJ, Sedivy JM. Telomere shortening triggers senescence of human cells through a pathway involving ATM, p53, and p21(CIP1), but not p16(INK4a) Mol Cell. 2004;14:501–513. [PubMed]
  • Hsu M, McEachern MJ, Dandjinou AT, Tzfati Y, Orr E, Blackburn EH, Lue NF. Telomerase core components protect Candida telomeres from aberrant overhang accumulation. Proc Natl Acad Sci U S A. 2007;104:11682–11687. [PMC free article] [PubMed]
  • Karlseder J, Broccoli D, Dai Y, Hardy S, de Lange T. p53- and ATM-dependent apoptosis induced by telomeres lacking TRF2. Science. 1999;283:1321–1325. [PubMed]
  • Kim M, Xu L, Blackburn EH. Catalytically active human telomerase mutants with allele-specific biological properties. Exp Cell Res. 2003;288:277–287. [PubMed]
  • Kirk KE, Blackburn EH. An unusual sequence arrangement in the telomeres of the germ-line micronucleus in Tetrahymena thermophila. Genes Dev. 1995;9:59–71. [PubMed]
  • Lansdorp PM, Verwoerd NP, van de Rijke FM, Dragowska V, Little MT, Dirks RW, Raap AK, Tanke HJ. Heterogeneity in telomere length of human chromosomes. Hum Mol Genet. 1996;5:685–691. [PubMed]
  • Li S, Rosenberg JE, Donjacour AA, Botchkina IL, Hom YK, Cunha GR, Blackburn EH. Rapid inhibition of cancer cell growth induced by lentiviral delivery and expression of mutant-template telomerase RNA and anti-telomerase short-interfering RNA. Cancer Res. 2004;64:4833–4840. [PubMed]
  • Masutomi K, Possemato R, Wong JM, Currier JL, Tothova Z, Manola JB, Ganesan S, Lansdorp PM, Collins K, Hahn WC. The telomerase reverse transcriptase regulates chromatin state and DNA damage responses. Proc Natl Acad Sci U S A. 2005;102:8222–8227. [PMC free article] [PubMed]
  • Mizuarai S, Yamanaka K, Kotani H. Mutant p53 induces the GEF-H1 oncogene, a guanine nucleotide exchange factor-H1 for RhoA, resulting in accelerated cell proliferation in tumor cells. Cancer Res. 2006;66:6319–6326. [PubMed]
  • Peng Y, Chen L, Li C, Lu W, Chen J. Inhibition of MDM2 by hsp90 contributes to mutant p53 stabilization. J Biol Chem. 2001;276:40583–40590. [PubMed]
  • Richards EJ, Chao S, Vongs A, Yang J. Characterization of Arabidopsis thaliana telomeres isolated in yeast. Nucleic Acids Res. 1992;20:4039–4046. [PMC free article] [PubMed]
  • Roninson IB. Tumor cell senescence in cancer treatment. Cancer Res. 2003;63:2705–2715. [PubMed]
  • Samper E, Flores JM, Blasco MA. Restoration of telomerase activity rescues chromosomal instability and premature aging in Terc−/− mice with short telomeres. EMBO Rep. 2001;2:800–807. [PMC free article] [PubMed]
  • Sarin KY, Cheung P, Gilison D, Lee E, Tennen RI, Wang E, Artandi MK, Oro AE, Artandi SE. Conditional telomerase induction causes proliferation of hair follicle stem cells. Nature. 2005;436:1048–1052. [PMC free article] [PubMed]
  • Sfeir AJ, Chai W, Shay JW, Wright WE. Telomere-end processing the terminal nucleotides of human chromosomes. Mol Cell. 2005;18:131–138. [PubMed]
  • Shay JW, Bacchetti S. A survey of telomerase activity in human cancer. Eur J Cancer. 1997;33:787–791. [PubMed]
  • Smith CD, Smith DL, DeRisi JL, Blackburn EH. Telomeric protein distributions and remodeling through the cell cycle in Saccharomyces cerevisiae. Mol Biol Cell. 2003;14:556–570. [PMC free article] [PubMed]
  • Taggart AK, Teng SC, Zakian VA. Est1p as a cell cycle-regulated activator of telomere-bound telomerase. Science. 2002;297:1023–1026. [PubMed]
  • Teixeira MT, Arneric M, Sperisen P, Lingner J. Telomere length homeostasis is achieved via a switch between telomerase- extendible and -nonextendible states. Cell. 2004;117:323–335. [PubMed]
  • Xu L, Blackburn EH. Human Rif1 protein binds aberrant telomeres and aligns along anaphase midzone microtubules. J Cell Biol. 2004;167:819–830. [PMC free article] [PubMed]
  • Zhu J, Wang H, Bishop JM, Blackburn EH. Telomerase extends the lifespan of virus-transformed human cells without net telomere lengthening. Proc Natl Acad Sci U S A. 1999;96:3723–3728. [PMC free article] [PubMed]
  • Zhu L, Hathcock KS, Hande P, Lansdorp PM, Seldin MF, Hodes RJ. Telomere length regulation in mice is linked to a novel chromosome locus. Proc Natl Acad Sci U S A. 1998;95:8648–8653. [PMC free article] [PubMed]
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...