Assembly of Mutant-Template Telomerase RNA into Catalytically Active Telomerase Ribonucleoprotein That Can Act on Telomeres Is Required for Apoptosis and Cell Cycle Arrest in Human Cancer Cells

The telomerase ribonucleoprotein is a promising target for cancer therapy, as it is highly active in many human malignancies. A novel telomerase targeting approach combines short interfering RNA (siRNA) knockdown of endogenous human telomerase RNA (hTer) with expression of a mutant-template hTer (MT-hTer). Such combination MT-hTer/siRNA constructs induce a rapid DNA damage response, telomere uncapping, and inhibition of cell proliferation in a variety of human cancer cell lines. We tested which functional aspects of the protein catalytic component of telomerase [human telomerase reverse transcriptase (hTERT)] are required for these effects using human LOX melanoma cells overexpressing various hTERTs of known properties. Within 3 days of MT-hTer/siRNA introduction, both growth inhibition and DNA damage responses were significantly higher in the setting of wild-type hTERT versus catalytically dead hTERT or mutant hTERT that is catalytically competent but unable to act on telomeres. These effects were not attenuated by siRNA-induced knockdown of the telomeric protein human Rap1 and were additive with knockdown of the telomere-binding protein TRF2. Hence, the effects of MT-hTer/siRNA require a telomerase that is both catalytically competent to polymerize DNA and able to act on telomeres in cells. (Cancer Res 2006; 66(11): 5763-71)

Telomerase-based therapy provides a novel approach to the treatment of many cancers. Telomeric DNA contains tandem repetitive DNA sequences located at the ends of human chromosomes (1). The two essential functions of telomeres are protecting chromosome ends (the “capping” function of telomeres) and facilitating their complete replication, usually involving the action of telomerase, a ribonucleoprotein enzyme. Telomerase consists of two core components: a reverse transcriptase protein [called human telomerase reverse transcriptase (hTERT) in humans] and human telomerase RNA (hTR or hTer in humans). hTer contains a short template sequence used by hTERT to synthesize telomeric DNA (2).

As a result of telomerase down-regulation in many normal somatic cells, human chromosomes can lose up to 50 to 200 nucleotides of telomeric sequence per cell division (3). Such shortening of telomeres has been proposed to be a mitotic clock that monitors cell divisions; sufficiently short telomeres and lack of telomerase may signal cellular senescence (3, 4). In contrast, highly active telomerase is a common feature of a wide variety of human cancers (5). Attempts to attenuate telomerase function in cultured human cancer cells have led to telomere shortening (6) and, eventually, to cellular apoptosis, thus limiting the cancer cell growth in vitro (7, 8) and providing further evidence that telomerase-dependent telomere maintenance promotes tumorigenesis. It was recently shown that depleting the endogenous wild-type hTer (wt-hTer) of cancer cells, with either short interfering RNAs (siRNA) or ribozymes, leads to reduced telomerase activity, growth inhibition, and reduction in metastatic progression in mouse models. Intriguingly, these effects of simply reducing telomerase RNA levels are accompanied by changes in the global gene expression profile, but they do not involve bulk telomere shortening or telomere uncapping nor do they elicit a DNA damage response or require p53 (9, 10).

A second telomerase-based strategy against cancer cells exploits the telomerase activation common in cancer cells, in effect turning telomerase itself into a weapon against these cells. This was accomplished by mutating the template of hTer and introducing that mutant-template hTer (MT-hTer) gene into cancer cells and mouse xenograft models (11, 12), where it resulted in cell killing and slowing of tumor growth. The effects were specific to the MT-hTer mutation and depended on telomerase expression but not on p53 or pRb function. However, in contrast to the effects of telomerase RNA depletion alone, MT-hTer expression induced a DNA damage response and formation of DNA damage foci, including DNA damage foci at telomeres (9, 13). Furthermore, combining siRNA targeting endogenous hTer together with MT-hTer on the same construct had a synergistic inhibitory effect on cell proliferation, consistent with more effective replacement of endogenous wt-hTer by MT-hTer (12). The construct containing both MT-hTer and siRNA against endogenous hTer will henceforth be termed MT-hTer/siRNA.

The hypothesis for the toxic effects of MT-hTer expression was that the MT-hTer complexed with hTERT to form a ribonucleoprotein that adds incorrect nucleotide repeats to the end of the telomere (14), preventing binding of telomere-protective proteins. However, whether it is the action of catalytically active telomerase on telomeres that causes the DNA damage and inhibition of proliferation in human cancer cells has remained unproven. An alternative mechanism of cell growth inhibition is suggested by studies of telomerase in yeast, in which wt-hTer has a direct telomere-protective function (15). This protection possibly involves the known in vivo association of telomerase with telomeres during G₁ phase (16, 17), a cell cycle stage during which polymerization by telomerase does not occur. These findings raised the possibility that an aberrant telomerase, such as that containing MT-hTer, may itself elicit a cellular response, for example, if it were unable to perform a comparable protective function in human cells, thus potentially contributing to the toxic effects of MT-hTer expression. Furthermore, the finding that siRNA directed against endogenous wt-hTer could alone inhibit cell growth (9) raised the possibility that the synergism observed between MT-hTer and the antitelomerase RNA siRNA might reflect other cause(s) besides the telomere uncapping predicted from incorporation of mutant telomeric DNA repeats.

Here, we test and investigate the mechanism by which MT-hTer/siRNA exerts its toxic effects. The experimental strategy was to introduce a MT-hTer/siRNA construct into cells that were already overexpressing one of three functional variants of the hTERT protein. The phenotypic effects of MT-hTer/siRNA were then determined in the setting of each hTERT variant. One hTERT mutant tested was a catalytically inactive hTERT enzyme, the “dominant-negative” point mutant D868A hTERT (DN-hTERT). The second hTERT mutant was N125A+T126A-hTERT, which creates an enzyme that is catalytically active in vitro but fails to elongate the telomeres in cells (18). The N125A+T126A mutation is located in the so-called “dissociates activities of telomerase” (N-DAT) region, thought to be necessary for action of telomerase on telomeres (19). Specifically, substitution of amino acids 122 to 127 in hTERT, which includes the two highly conserved residues N125 and T126, has been shown previously to interfere with the efficient utilization of telomeric G-rich DNA primer substrates by reconstituted telomerase in vitro, although the ability to elongate a nontelomeric DNA substrate remained the same as wild-type (20). Hence, the N125 and T126 residues are thought to be involved in the “anchor” site function of telomerase that increases the affinity of telomerase for its telomeric DNA primer substrate to be elongated.

The results of the experiments described here indicate that, to exert its toxic effects, MT-hTer/siRNA needs to assemble into a telomerase enzyme complex containing a hTERT that is both catalytically competent and can act on telomeres in the cell and that the potency of the rapid cell growth inhibition is not weakened by depletion of telomeric proteins TRF2 or human Rap1 (hRap1).

Plasmid construction. The hTERT alleles were engineered as described previously (18). Briefly, the N125A+T126A mutations at codons 125 and 126 and the D868A mutation at codon 868 were introduced using PCR-based site-directed mutagenesis. All of the hTERT alleles were cloned into a pBabe-puro retroviral expression vector. The MT-hTer/siRNA was engineered as described previously (12). Briefly, wt-hTer was PCR cloned from human genomic DNA and subcloned into BlgII/SalI site in pIU1-T7 plasmid (generously provided by Dr. Edouard Bertrand, University Montpellier, Montpellier, France). MT-hTer was then generated by site-directed mutagenesis to create an hTer with the templating domain sequence 5′-UAUAUAUAUAA-3′, AU5-MT-hTer (11, 12). The siRNA targeting wt-hTer was engineered by PCR using a pTZ-U6 plasmid (generously provided by Dr. John J. Rossi, Beckman Research Institute of the City of Hope, Duarte, CA) using primers as described previously (12). The PCR product was subcloned along with the MT-hTer into pHRCMVGFPWSin18 vector. siRNA targeting TRF2 and hRap1, as well as a scrambled siRNA sequence that does not affect mRNA levels or cell proliferation, was packaged using this same lentiviral vector system with sequences as described previously (13).

Virus production. Recombinant lentiviruses were generated by transfecting pMD.G and pCMVDR8.91 plasmids (generously provided by Dr. Didier Trono, University of Geneva, Geneva, Switzerland; ref. 21), along with the MT-hTer/siRNA plasmid, into 293T cells with LipofectAMINE 2000 (Invitrogen, Carlsbad, CA). Recombinant retroviruses were generated by transfecting hTERT pBabe-puro plasmids into Phoenix packaging cell line with LipofectAMINE 2000. For both lentiviral and retroviral transfections, virus-containing medium was harvested 48 and 72 hours after transfection and filtered through 0.45-μm filters.

Cell culture and infection. LOX melanoma cells were seeded in six-well tissue culture plates at 2 × 10⁴ per well in duplicate or triplicate on the evening before infection. Infection with pBabe-puro retrovirus containing WT-hTERT or DN-hTERT or N125A+T126A-hTERT was achieved by incubation with virus-containing culture medium supplemented with 8 μg/mL polybrene for 8 hours. After 8-hour infection, fresh culture medium was added, and the cells were maintained in culture for 2 days. At the end of day 2, puromycin-containing culture medium was added, and the cells were maintained in culture for 4 more days. At the end of 4 days of puromycin selection (day 6 after hTERT infection), cells were reseeded at 2 × 10⁴ per well in duplicate. On the following day (day 7 after hTERT infection), cells were infected with lentiviral MT-hTERT/siRNA constructs at >95% efficiency of infection as indicated by green fluorescent protein (GFP) expressed from the same lentiviral vector. Lentiviral siRNA constructs targeting TRF2 or hRap1 were infected in a similar fashion. Cells were counted at 3, 6, and 9 days after lentiviral infection using a hemocytometer under direct illumination microscopy.

Apoptosis measurement by fluorescence-activated cell sorting. After cells were trypsinized and counted at each time point, 1 × 10⁵ cells were washed and resuspended in PBS and then stained with Annexin-allophycocyanin conjugate (Invitrogen) per manufacturer's protocol. Cells were also stained with 1 μg/mL propidium iodide (Invitrogen) as a dead cell counterstain. Stained cells were fluorescence sorted on a FACSAria cell sorter (Becton Dickinson, San Jose, CA).

Cell cycle analysis by fluorescence-activated cell sorting. Immediately before trypsinizing and counting the cells at each time point, a subset of cells was treated with 10 μmol/L BrdUrd (Sigma, St. Louis, MO) for 1 hour. Trypsinized BrdUrd-treated cells were washed in PBS and fixed with 70% ethanol. Cells were stained overnight using anti-BrdUrd Alexa Fluor 647 monoclonal antibody (Invitrogen) per manufacturer's protocol and then fluorescence sorted on a FACSCalibur cell sorter (Becton Dickinson).

Immunofluorescence microscopy. Cells were grown on glass coverslips for 3 days, fixed with 2% formaldehyde in PBS, and permeabilized with 0.2% Triton X-100 in PBS. Immunostaining was done using primary antibodies against 53BP1 (BD Transduction Laboratories, San Jose, CA) followed by secondary antibody conjugated with Alexa Fluor 568 (Invitrogen). DNA was visualized with 0.2 μg/mL 4′,6-diamidino-2-phenylindole (DAPI). All images were analyzed using a Nikon (Kawasaki, Japan) Eclipse E600 microscope with 60× or 100× objective.

MT-hTer alone causes early cell growth inhibition, apoptosis, and cell cycle arrest. Before the experiments combining expression of MT-hTer/siRNA and hTERT variants (see Materials and Methods; Fig. 1), separate experiments were done to characterize and control for the timing and magnitude of growth inhibition contributed by each individual construct. First, LOX melanoma cells (a telomerase-positive human cancer cell line) were infected separately with lentiviral vectors expressing either the MT-hTer or the siRNA directed at wt-hTer (Fig. 1A and B). Both constructs contained a GFP marker, and infection efficiency was >95% as quantified by percentage of cells with GFP fluorescence. MT-hTer expression and knockdown of wt-hTer by the siRNA were confirmed by Northern and reverse transcription-PCR analyses (Supplementary Figs. S1 and S2). Growth analysis revealed that although both constructs exerted an inhibitory effect, MT-hTer produced marked growth inhibition in this cell line almost immediately after its introduction, readily measurable by day 3 and statistically significant by day 6 (P = 0.02; Fig. 2A). In contrast, in this experiment, siRNA directed against wt-hTer only began to inhibit cell growth after ∼1 week from its introduction; although slowing of cell proliferation was detected by day 9, in these experiments, the difference by day 9 had not achieved statistical significance.

To characterize the mechanism of growth inhibition induced by MT-hTer expression, at each time point, cells were quantitatively assessed for apoptosis and cell cycle arrest using Annexin V and BrdUrd staining, respectively (Fig. 2B and C). In the cells overexpressing MT-hTer, a 3-fold increase in apoptosis was apparent by day 3 and remained statistically significant (P < 0.05) throughout the experiment (Fig. 2B). At day 3, MT-hTer expression also led to a 28% relative reduction in S-phase fraction (from 46% to 33% in S phase) and a concomitant increase in G₂ fraction, suggestive of a G₂ arrest (Fig. 2C). The reduction in S-phase fraction induced by MT-hTer was less pronounced at the later time points possibly because the BrdUrd analysis was done on unselected populations, which over time became enriched for cells that were not expressing the growth-inhibitory MT-hTer. In contrast to the early effects of MT-hTer, no significant increase in apoptosis or cell cycle arrest was detectable in cells overexpressing just the siRNA against wt-hTer for up to 9 days. Notably, control LOX cells infected with empty vector only showed a level of baseline apoptosis. This phenomenon was not attributable to the lentiviral infection, as we observed similar levels of baseline apoptosis in uninfected control LOX cells using the same Annexin V assay (data not shown). Thus, we attribute this finding to the baseline characteristic of the assay in this cell line.

MT-hTer/siRNA inhibits proliferation in the setting of WT-hTERT but is minimally effective in the setting of catalytically inactive hTERT or a DAT mutant hTERT. Having characterized the growth inhibition effects of each individual construct, we proceeded to investigate the mechanism of action of MT-hTer/siRNA. The MT-hTer/siRNA was introduced into LOX cells, in which either WT-hTERT or one of two mutant hTERT alleles of known properties (18) was ectopically overexpressed (Fig. 1C). WT-hTERT was used as a control for ectopic overexpression of the catalytic subunit of telomerase, hTERT. The first hTERT mutant was a catalytically inactive hTERT enzyme, the DN-hTERT. The second hTERT mutant was N125A+T126A-hTERT, which creates a DAT mutant enzyme that is catalytically active in vitro but fails to elongate the telomeres (18, 20).

To ensure that the effects of the MT-hTer/siRNA construct could be compared in cells that were growing at equal rates, the short-term effect on LOX cell growth of overexpressing each of the hTERT alleles alone was characterized. LOX cells were infected separately with retroviral vectors expressing WT-hTERT, DN-hTERT, or N125A+T126A-hTERT. Equivalent levels of expression for each of the hTERT variants were confirmed by Northern analysis (Supplementary Fig. S3). Previously published work has shown that the D868A (DN) and N125A+T126A (DAT) mutations do not affect the ability of hTERT to assemble with hTer (22, 23) and that ectopic expression of the various hTERT alleles produces telomerase ribonucleoprotein with the expected enzymatic activity (18, 19, 22). This was confirmed in the present experiments using LOX cells by telomeric repeat amplification protocol (TRAP) assays, showing preserved telomerase activity in the WT-hTERT and N125A+T126A-hTERT mutants and diminished activity (80-85% reduction) in the DN-hTERT mutant (Supplementary Fig. S4). After the LOX cells were retrovirally infected with the hTERT variants, growth analysis of the puromycin-selected cell populations revealed no effect on proliferation for up to 14 days by any of the hTERT alleles (Fig. 3). Thus, although overexpression of the mutant hTERTs eventually leads to telomere shortening and has long-term effects on cell growth (12, 18), there were no growth rate effects seen in the duration of the present experiments.

The requirements for MT-hTer/siRNA activity were tested with a serial infection experiment (Fig. 1D). First, retroviral vectors containing each of the hTERT alleles were introduced into LOX cells and subjected to puromycin selection as above. Then, lentiviral vector containing MT-hTer/siRNA and GFP, or an empty control vector containing GFP only, was introduced into those same LOX cells at high efficiency as quantified by percentage of cells with GFP fluorescence. The DNA damage and cell growth responses to the MT-hTer/siRNA were thus quantified in the settings of overexpressed WT-hTERT or overexpressed mutant hTERTs. All infections were done in duplicate, and the experiment was repeated with similar results.

LOX cell proliferation was quantified using hemocytometry on days 3, 6, and 9 after MT-hTer/siRNA was introduced (Fig. 4A and B). By day 3, cell proliferation was significantly inhibited (P = 0.001) in the setting of WT-hTERT expression. In contrast, cells expressing DN-hTERT or N125A+T126A-hTERT were not significantly affected by MT-hTERT/siRNA at days 3 or 6, although, on day 9, MT-hTer/siRNA-induced growth inhibition in the DN-hTERT cells became statistically significant. Overall, on days 3, 6, and 9, MT-hTer/siRNA induced progressively greater growth inhibition,∼3-, 13-, and 19-fold, respectively, in the setting of WT-hTERT expression. In contrast, cells expressing DN-hTERT or N125A+T126A-hTERT experienced no growth inhibition at day 3 after introduction of MT-hTer/siRNA, only 1.5-fold inhibition at day 6 (not statistically significant), and ∼3.5-fold at day 9 [showing statistical significance (P = 0.03) for DN-hTERT only; Fig. 4B and C].

MT-hTer/siRNA induces DNA damage foci in the setting of WT-hTERT but elicits significantly fewer such foci in the presence of catalytically inactive hTERT or a DAT mutant hTERT. It was previously shown that the growth inhibition induced by two different MT-hTer/siRNA constructs, the AU5 construct used in the present experiments and 47A MT-hTer, was associated with the appearance of DNA damage foci in treated cells (9, 13). For 47A, >80% of the foci colocalized at telomeres; in the AU5-MT-hTer cells, a lower percentage of foci were telomeric (9, 13).³

To quantify DNA damage focus formation in the AU5-MT-hTer/siRNA-treated cells, at each time point (days 3, 6, and 9), in parallel to the cell counts described above, a subset of cells was grown on glass coverslips and examined microscopically following fixation and staining with antibody against 53BP1, which localizes to DNA damage foci (13). In the LOX melanoma cells expressing MT-hTer/siRNA in the setting of WT-hTERT (Fig. 5A), large numbers of bright, prominent 53BP1 foci appeared, consistent with DNA damage foci as shown previously (9, 13). Costaining with DAPI (Fig. 5A,, blue) showed that these foci localized to the nuclei, and GFP positivity (Fig. 5A,, green) indicated that the cells expressed the MT-hTer/siRNA. Cells overexpressing MT-hTer/siRNA in the setting of WT-hTERT also displayed typical morphologic changes characterized by overall enlargement and rounding of cell bodies, relative loss of cytoplasmic projections, nuclear enlargement and multilobulation, cytoplasmic vacuolization, and occasional presence of anaphase bridges (Fig. 5A).

For each hTERT variant (WT-hTERT, DN-hTERT, or N125A+T126A-hTERT) infected with MT-hTer/siRNA or vector control, we counted 100 cells in duplicate and scored each cell as “53BP1 positive” (≥1 53BP1 foci) or negative (Fig. 5B). LOX cells expressing WT-hTERT had significantly more 53BP1 DNA damage foci than those expressing DN-hTERT or N125A+T126A-hTERT: on day 3, 42% of WT-hTERT cells that received MT-hTer/siRNA contained one or more distinct 53BP1 foci compared with 20% to 22% of the DN-hTERT and N125A+T126A-hTERT cells that received MT-hTer/siRNA (P = 0.002 and 0.01, respectively) and 15% to 20% of any hTERT cells that received vector control. In a repeat experiment, by day 3, 20% of WT-hTERT cells that received MT-hTer/siRNA contained multiple (≥4) 53BP1 foci, compared with only 2% to 4% in the DN-hTERT and N125A+T126A-hTERT cells that received MT-hTer/siRNA (data not shown). Hence, in the setting of WT-hTERT, the production of DNA damage foci induced by MT-hTer/siRNA correlated with growth inhibition.

MT-hTer inhibition of proliferation is not attenuated by knockdown of other telomere-associated proteins. The mechanism by which the addition of mutant template to telomeres inhibits cell growth was investigated further, first by knocking down the telomeric protein TRF2, which binds telomeric DNA sequence specifically and is involved in normal telomere end protection and length regulation (24, 25). The hairpin siRNA used was the same siRNA construct recently validated in our laboratory to knock down TRF2 protein levels in the same LOX cell line used in the present work (13). This siRNA construct was expressed in the LOX cells using the same lentiviral infection protocol described previously (13). TRF2 knockdown after siRNA expression in LOX cells was confirmed by Western immunoblot (Supplementary Fig. S5A). Three days after infection with the anti-TRF2 siRNA, a second infection with a lentiviral construct containing just the MT-hTer was done. Consistent with successful knockdown of TRF2, a significant reduction in cell proliferation, to ∼5 × 10⁵ down from ∼9.5 × 10⁵ in control samples treated in parallel with scrambled siRNA, was observed (Fig. 6A,, blue columns). Expression of the MT-hTer on top of this TRF2 knockdown further inhibited cell growth down to 2.8 × 10⁵ cells (P = 0.02; Fig. 6A , maroon columns).

A similar experiment was done in which the levels of the protein hRap1 were knocked down. hRap1 is recruited by TRF2 to the telomere, where it acts as a negative regulator of telomere length (26). The present knockdown experiments were done again using the same hairpin siRNAs recently validated in our laboratory to knock down hRap1 in the same LOX cell line (13), and knockdown was confirmed by Western immunoblot (Supplementary Fig. 5SB). In contrast to the above findings with siRNAs targeting TRF2, the short-term knockdown of hRap1 alone did not significantly reduce cell proliferation (Fig. 6B,, blue columns). Expression of the MT-hTer in these cells caused significant growth inhibition as in the control cells that received the scrambled siRNA (P = 0.04; Fig. 6B , maroon columns). Therefore, depleting hRap1 did not diminish the effect of the MT-hTer on cell growth.

A telomerase-targeting construct has previously been developed that is highly toxic to a variety of cancer cell lines, inducing profound growth inhibition, DNA damage, and apoptosis over a time course of days (12). This construct was designed to take advantage of a two-pronged approach, siRNA that knocks down endogenous wt-hTer coupled with a MT-hTer. The results reported here are consistent with the interpretation that the effects of MT-hTer/siRNA require mutant telomeric DNA synthesis on telomeres. These results provide the most direct evidence to date that the mechanism by which MT-hTer/siRNA exerts its toxic effects on human cancer cells is to combine with endogenous hTERT, forming a catalytically active telomerase that is capable of adding incorrect DNA repeats to the telomeres.

We showed this mechanism of action of MT-hTer/siRNA by a novel approach using telomerase enzyme (hTERT) variants with previously characterized properties. We began with the following hypothesis. Two conditions must both be met in order for the MT-hTer-hTERT ribonucleoprotein to exert its toxic effects by the proposed mechanism (adding incorrect nucleotides to the telomere): (a) the hTERT must be enzymatically active and (b) the hTERT must be able to act on the telomeres in vivo. We tested this hypothesis by ectopically overexpressing three hTERT alleles before treatment with the MT-hTer/siRNA construct: functional WT-hTERT, enzymatically inactive DN-hTERT, and the DAT N125A+T126A-hTERT. Then, we quantified the effects of MT-hTer/siRNA in each of these intracellular hTERT milieus.

Before the serial infection experiments, we quantified the individual effects on LOX cell proliferation contributed by each of the component constructs comprising MT-hTer/siRNA. When tested separately, the MT-hTer produced cell cycle arrest, apoptosis, and consequent growth inhibition significantly more rapidly and intensely than the siRNA directed against wt-hTer. The independent growth-inhibitory effects of siRNA directed at wt-hTer were not significant over the time course of the present experiments. These results suggest that, in this cell line, the earliest phase of the cancer cell inhibition by MT-hTer/siRNA is primarily dependent on the mutant template, likely through its previously shown ability to induce a DNA damage response (12). Notably, recent evidence also indicates that siRNA directed at wt-hTer functions via additional pathways (9). Ongoing studies in our laboratory aim to separate and better define the respective contributions of MT-hTer and siRNA against wt-hTer to the phenotypes of DNA damage and growth inhibition.

In separate experiments, the WT-hTERT-overexpressing, DN-hTERT-overexpressing, and N125A+T126A-hTERT-overexpressing LOX cells all grew similarly for the short periods (10-14 days) monitored after selection. Because overexpression of each of the hTERT constructs in LOX cells did not significantly alter the short-term LOX growth phenotype, we were able to quantify any cell growth differences arising from subsequent expression of MT-hTer/siRNA against a constant cell proliferation rate. Hence, the differential effects observed when MT-hTer interacted with each of the hTERT alleles can be attributed to the differences in those predefined functional capacities of hTERTs rather than to some other inherent differences between the hTERT alleles. This was a critical validation, especially in light of recent publications showing a significant long-term effect on cell activation produced by TERT expression alone in some systems (27, 28).

When we proceeded to introduce the combined MT-hTer/siRNA construct into cells expressing the hTERT alleles, as early as 3 days after introduction of MT-hTer/siRNA, there was a marked inhibition of proliferation in the setting of WT-hTERT expression (>2-fold reduction in cell number) but no growth inhibition in cells expressing the DN-hTERT or N125A+T126A-hTERT alleles. These growth effects correlated with the results of 53BP1 staining, which showed that the proportion of cells sustaining any demonstrable DNA damage as a result of MT-hTer/siRNA expression was ∼2- to 3-fold higher in the WT-hTERT-expressing cells than in the mutant hTERT-expressing cells. Similarly, the proportion of cells with multiple (≥4) foci of DNA damage was ∼6-fold higher in the WT-hTERT-expressing cells than in the mutant hTERT-expressing cells.

These data show that MT-hTer/siRNA causes DNA damage and growth inhibition in a manner dependent on the action of WT-hTERT, with MT-hTer/siRNA losing effectiveness if the introduced hTERT is enzymatically inactive (DN-hTERT). Similarly, MT-hTer/siRNA was ineffective in the setting of a DAT hTERT allele that has lost the ability to elongate telomeres in vivo (N125A+T126A-hTERT). Previous publications have shown that mutations in the DAT region render hTERT specifically unable to act on telomeric primers (20, 29) and that in vivo activity could be restored to the DAT mutant through fusion to TRF2, which localizes to the telomere (30). Thus, the relative absence of MT-hTer/siRNA growth-inhibitory effects in the setting of the N125A+T126A-hTERT mutant, which has an impaired capacity to act on telomeres, signifies that the MT-hTer/siRNA toxic effects most likely occur at the telomere. This is consistent with previous findings that DNA damage foci induced by MT-hTer expression colocalize at telomeres, with up to 80% telomeric colocalization observed with some constructs (9, 13).³ Therefore, we conclude that MT-hTer must complex with enzymatically active hTERT and act on telomeres to exert its effects. The most likely explanation for these results is that MT-hTer/siRNA does in fact work through addition of incorrect nucleotides to the telomere and a DNA damage response, leading to inhibition of growth.

Although MT-hTERT/siRNA was significantly more toxic to the cells expressing WT-hTERT, later in the time course, we also noted a modest growth-inhibitory trend in the cells expressing DN-hTERT and N125A+T126A-hTERT, which was not statistically significant, except with DN-hTERT at day 9. We speculate that, occasionally, the N125A+T126A-hTERT can act on telomeres or that endogenous hTERT, although expected generally to be outcompeted by the overexpressed mutant forms, assembled with the MT-hTer. This possibility is born out by the TRAP assay analysis of the hTERT variants (Supplementary Fig. S4), which showed that telomerase activity in the DN-hTERT-expressing cells, though diminished by 80% to 85% relative to parental controls, was still present. Another contributing factor may be that the siRNA portion of the construct may begin to exert its independent growth-inhibitory effects by day 9 as suggested by the growth results shown in Fig. 2A and in previous work (9, 12). Any such late inhibitory effects of the siRNA are expected to be more readily discernable in the cells that expressed DN-hTERT or N125A+T126A-hTERT because, unlike the WT-hTERT-expressing cells, the DN-hTERT-expressing and N125A+T126A-hTERT-expressing cells were relatively unaffected by the earlier, more potent growth inhibition caused by the MT-hTer portion of the construct.

We tested whether the effects of MT-hTer/siRNA were mediated by the telomere-interacting proteins hRap1 and TRF2. hRap1 does not bind normal telomeric DNA directly; nevertheless, its involvement at the telomere is substantiated by a growing body of research. In budding yeast, Rap1 binds telomeric DNA directly and acts as a negative regulator of telomere length (31, 32). The function of Rap1 seems to be conserved in its mammalian orthologue, hRap1, which is recruited by TRF2 to the telomere, where it acts as a negative regulator of telomere length (26). The present experiments with hRap1 sought to define its role in the growth inhibition generated by MT-hTer. MT-hTer expression produced the same degree of growth inhibition in hRap1 knockdown cells as in control cells. Hence, hRap1 depletion does not reduce the effect of MT-hTer in inducing growth inhibition.

TRF2 directly binds telomeric DNA sequence specifically and also recruits hRap1 to the telomeres (24, 25). It has been proposed that incorporation of aberrant telomeric repeats may disrupt the normal binding of TRF2 to the telomeric DNA (11, 33). Disruption of TRF2 binding, in turn, may lead to telomere “uncapping” and a DNA damage response as suggested by one set of experiments in HeLa cancer cells, where expression of a dominant-negative TRF2 allele that lacked the telomeric DNA binding domain and depleted TRF2 from telomeres led to increased apoptosis in a p53-dependent manner (34). In the present experiments, TRF2 knockdown alone produced the expected growth inhibition phenotype in LOX cells as described previously (13). Notably, expression of MT-hTer in these cells after TRF2 had been knocked down led to a further reduction in cell proliferation rate. Relative to vector control, MT-hTer expression produced a degree of growth inhibition comparable to that seen when the MT-hTer was expressed with no TRF2 knockdown. Therefore, the relative absence of TRF2 seems to act in an additive fashion with MT-hTer. Interestingly, previous work in our laboratory has shown that the apoptotic response generated by MT-hTer in HCT116 cancer cells or a transformed fibroblast cell line is not dependent on p53 (12). Hence, TRF2 depletion and MT-hTer effects may involve different pathways.

In summary, the current studies offer insight into a phenomenon that has been reported repeatedly (11, 12, 35), that interference with telomerase elicits an acute apoptotic response that cannot be explained by the traditional role ascribed to telomerase (prevention of gradual telomere shortening with serial cell divisions). Here, we show that DNA damage, cellular apoptosis, and inhibition of cancer cell proliferation indeed occur rapidly when telomerase function is disrupted by introduction of MT-hTer but only when that MT-hTer incorporates into a catalytically active telomerase capable of normal action on telomeres in vivo. These findings, together with the additive effects of growth inhibition caused by knockdown of the telomere-interacting proteins hRap1 and TRF2, suggest that MT-hTer exerts its toxicity by altering critical interactions at the telomere. Thus, our work begins to elucidate not only the manner by which a novel agent, MT-hTer, inhibits cancer cell proliferation but also the underlying delicate balance that makes telomerase interference a promising therapeutic approach.

Grant support: NIH Basic Research in Hematology and Oncology T32 grant DK007636 (A. Goldkorn) and NIH/National Cancer Institute grant CA96840 and Bernard Osher Foundation (E.H. Blackburn).

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.

We thank Jue Lin and Dana Smith for critical reading of the article, Maura Devlin-Clancy for helping with the preparation and editing of the article, and Brad Stohr for helpful input on experiment design.