• Open Access

Mitochondrial hTERT exacerbates free-radical-mediated mtDNA damage


  • Janine Hertzog Santos,

    1. Laboratory of Molecular Genetics,
    Search for more papers by this author
  • Joel N. Meyer,

    1. Laboratory of Molecular Genetics,
    Search for more papers by this author
  • Milan Skorvaga,

    1. Laboratory of Molecular Genetics,
    2. Laboratory of Molecular Carcinogenesis, Cancer Research Institute, Slovak Academy of Sciences, Vlarska 7, 833 91 Bratislava, Slovak Republic
    Search for more papers by this author
  • Lois A. Annab,

    1. Laboratory of Molecular Carcinogenesis, National Institute of Environmental Health Sciences, National Institutes of Health, 111, Alexander Drive, MD D3-01, Research Triangle Park, NC 27709, USA, and
    Search for more papers by this author
  • Bennett Van Houten

    Corresponding author
    1. Laboratory of Molecular Genetics,

      Dr Bennett Van Houten, Laboratory of Molecular Genetics – NIEHS, 111 Alexander Drive, PO Box 12233, Research Triangle Park, NC 27709, USA. Tel.: +1 919 541 7752; fax: +1 919 541 7593; e-mail: vanhout1@niehs.nih.gov
    Search for more papers by this author

Dr Bennett Van Houten, Laboratory of Molecular Genetics – NIEHS, 111 Alexander Drive, PO Box 12233, Research Triangle Park, NC 27709, USA. Tel.: +1 919 541 7752; fax: +1 919 541 7593; e-mail: vanhout1@niehs.nih.gov


Telomerase is often re-activated in human cancers and is widely used to immortalize cells in culture. In addition to the maintenance of telomeres, telomerase has been implicated in cell proliferation, genomic instability and apoptosis. Here we show that human telomerase reverse transcriptase (hTERT) is targeted to the mitochondria by an N-terminal leader sequence, and that mitochondrial extracts contain telomerase activity. In seven different human cell lines, mitochondrial telomerase increases hydrogen-peroxide-mediated mitochondrial DNA damage. hTERT expression did not alter the rate of hydrogen peroxide breakdown or endogenous cellular levels. Because the damaging effects of hydrogen peroxide are mediated by divalent metal ions (Fenton chemistry), we examined the levels of bioavailable metals. In all cases, higher levels of chelatable metals were found in hTERT-expressing cells. These results suggest that mitochondrial telomerase sensitizes cells to oxidative stress, which can lead to apoptotic cell death, and imply a novel function of telomerase in mitochondrial DNA transactions.


Normal human cells have a limited replicative lifespan when propagated in vitro (Hayflick & Moorehead, 1961). This is believed to be due to the shortening of telomeres that occurs each time normal human cells divide. Telomeres are specialized DNA–protein structures at the end of eukaryotic chromosomes that act to preserve chromosome integrity by preventing degradation, end-to-end fusions, rearrangements and chromosome attrition (Greider, 1991; Blackburn, 1994). Primary human fibroblasts undergo progressive shortening of telomeres, resulting in replicative senescence at 60–80 cell divisions (Harley et al., 1990; Hayflick & Moorehead, 1961). However, expression of telomerase – a ribonucleoprotein that synthesizes telomeric repeats – not only stabilizes telomeres but also prevents replicative senescence. More importantly, it endows cells with unlimited proliferative potential (Bodnar et al., 1998; Vaziri & Benchimol, 1998).

Human telomerase has two essential components, a catalytic protein subunit (human telomerase reverse transcriptase, hTERT) and a small RNA molecule (hTR) containing the template used to synthesize new telomeric repeats (Collins, 2000). Whereas the latter is widely expressed in embryonic and somatic tissue, hTERT is tightly regulated and is not detectable in most somatic cells (Weinrich et al., 1997). Despite this fact, highly proliferative cells such as germline cells, trophoblasts, haematopoietic cells, endometrial cells and up to 85% of cancer cells express telomerase to varying degrees (Counter et al., 1994a,b, 1995; Kim et al., 1994; Wright et al., 1996; Yasumoto et al., 1996; Shay & Bacchetti, 1997). Recent evidence suggests that telomerase may be transiently expressed during the S-phase of normal human cells (Masutomi et al., 2003).

The ectopic expression of telomerase in somatic cells was shown to result in telomere elongation and indefinite lifespan (Bodnar et al., 1998). Notably, it was also demonstrated that hTERT expression does not lead to neoplastic transformation of cultured cells (Morales et al., 1999).

A number of studies have investigated the consequences of hTERT expression in relation to various physiological processes, including: cell proliferation (Buchkovich, 1996; Greider, 1998; Gorbunova et al., 2002; Masutomi et al., 2003; Ramirez et al., 2002); apoptosis (Gorbunova et al., 2002; Ramirez et al., 2003); and cancer (Avilion et al., 1996; Greider, 1998; Cerni, 2000; Hackett & Greider, 2002; Artandi et al., 2003; Stewart et al., 2002). The effects of hTERT on other biological functions, such as mitochondrial physiology, have not been reported. We recently demonstrated that H2O2-induced mitochondrial DNA (mtDNA) damage is partially repaired after 6 h in normal human fibroblasts (NHF) transfected with hTERT. After this recovery time, repair activity was no longer detected in the mitochondria due to loss of the mitochondrial inner membrane potential. Under the same conditions, the nuclear DNA (nDNA) of these cells seemed completely resistant to H2O2-induced damage (Santos et al., 2003). During the course of these studies we noted that the mtDNA of NHF hTERT cells suffered more H2O2-induced damage as compared with other cell lines previously used in our laboratory. The present study explores the possible connection between hTERT expression and increased susceptibility of human cells to oxidative stress. By using confocal microscopy and an hTERT–EGFP fusion protein, we show that hTERT is present in the mitochondria. Moreover, hTERT presence in this organelle increases the levels of H2O2-induced mtDNA damage. Finally, we demonstrate that the effects of hTERT on the mtDNA appear to be, in part, due to its modulation in the levels of bioavailable divalent metals that can undergo Fenton chemistry.


hTERT is localized to the mitochondria

Our previous findings suggested that ectopic hTERT may increase the damaging effects of H2O2 in the mitochondria of human cells (Santos et al., 2003), and led us to hypothesize that hTERT might act directly in the mitochondria. Protein import into the mitochondria frequently relies on the presence of a targeting signal that is usually contained within the N-terminus of the protein (presequence), which is cleaved upon import into the organelle. The basic features of a premitochondrial targeting sequence are: (i) 20–60 amino acids with abundant positive charges, (ii) very few if any residues with negative charges, (iii) frequent hydroxylated amino acids and (iv) tendency to fold in an amphiphilic α-helix (Neupert, 1997). Different software packages are available to predict mitochondrial localization of a given protein. We used TargetP V1.0 and MitoProt II V1.0a4 to assess whether hTERT has a predicted mitochondrial-localization polypeptide. Unexpectedly, the human TERT gene showed a strong likelihood of having a mitochondrial leader sequence with a predicted cleavage site of 20 amino acids (see Fig. 1). Interestingly, alignment of the sequence of the TERT gene of several species showed that this sequence is present only in the gene of higher eukaryotes such as plants and mammals (Fig. 1).

Figure 1.

hTERT premitochondrial sequence. Alignment of the amino acid sequence of the TERT gene of various organisms; the mitochondria targeting sequences are shaded in gray.

In order to confirm that the putative mitochondrial leader sequence of hTERT targets it to the mitochondria, we created a full-length hTERT–EGFP fusion protein, and transiently transfected this construct into NHF primary fibroblasts, as well as in HeLa cells (which express endogenous telomerase). Mitotracker red was used as a mitochondrial probe. As previously documented, hTERT is clearly observed in the nucleus (Fig. 2C,G). However, a fraction of the protein is consistently seen in the mitochondria, as indicated by co-localization of the full-length hTERT–EGFP fusion with the Mitotracker-red-stained organelles (Fig. 2D,H). In fact, all the cells analysed that had the EGFP-fused protein showed the presence of hTERT in the mitochondria. These results give compelling evidence that, in addition to being in the nucleus, hTERT is also present in the mitochondria. It was not possible to monitor localization of endogenous hTERT using immunofluorescence because of the lack of specificity of the antibodies against hTERT that are commercially available (data not shown).

Figure 2.

Mitochondrial localization of hTERT. NHF (upper panels) and HeLa cells (bottom panels) were stained with Mitotracker red (A and E), and transiently transfected with EGFP-N1 alone (inset), or with the full-length hTERT–EGFP fusion (C and G). Cells were analysed 24 h after transfections and co-localization of hTERT–EGFP and mitochondria analysed (D and H). Panels B and F show phase contrast images.

To ensure that the predicted mitochondrial leader sequence of hTERT is in fact what is targeting this enzyme to the mitochondria, we created another EGFP construct harbouring a fusion of the first 20 amino acids of hTERT to the N-terminus of the fluorescent protein (20aahTERT–EGFPpCMV-myc). The sequence of amino acids cloned is shown in Fig. 1 (highlighted in gray). It is clear from the confocal images that 20aahTERT–EGFPpCMV-myc localizes strictly to mitochondria (Fig. 3E,F) and that the intracellular distribution of the fusion protein is different from that seen in the EGFPpCMV-myc only (Fig. 3B,C). The same results were obtained when 20aahTERT–EGFP was cloned onto the pCMV-HA vector (data not shown). Taken together, these two experimental results demonstrate that the N-terminus of hTERT has a true mitochondrial leader sequence.

Figure 3.

hTERT has a true mitochondrial leader sequence. The first 20 amino acids of the putative N-terminal mitochondrial leader sequence of hTERT were cloned on the N-terminus of EGFP. This fusion protein was subsequently subcloned into the pCMV-myc vector and transiently transfected in HeLa cells. Twenty-four hours after transfection, cells were analysed under a confocal microscope for subcellular localization. Mitochondria were stained with 5 nm of mitotracker red for 5 min. Upper panels show images of EGFP-pCMV-myc transfected cells: (A) mitotracker red only; (B) EGFP only; (C) merging of both images. Lower panels show cells transfected with 20aahTERT-fused to the EGFP protein: (D) mitotracker red only; (E) 20aahTERT–EGFP; (F) merging of both images.

Active telomerase is present in the mitochondria of human cells

In order to verify whether hTERT present in the mitochondria results in active telomerase, three differently derived fibroblast strains (NHF, MRC-5 and NHF LXIN) were stably transfected with a cassette containing the full-length hTERT gene by retroviral infection (NHF hTERT, MRC-5 hTERT, NHF LXIN hTERT). Depending on the integration site, hTERT transfections can lead to variable levels of telomerase protein expression. Therefore, we first measured the levels of telomerase activity in total protein extracts of the cells described above using the telomeric repeat amplification protocol (TRAP) (Kim et al., 1994). The TRAP assay demonstrated that MRC-5 hTERT has the lowest telomerase activity followed by NHF hTERT and NHF LXIN hTERT. Additionally, none of the isogenic primary parentals shows telomerase function (Fig. 4A).

Figure 4.

Variable levels of telomerase activity in the nucleus and in the mitochondria of hTERT-expressing cells. (A) Telomerase activity was monitored with the TRAP assay. The activity of the enzyme was examined from 750 ng of total cellular extracts. As a control for specificity of the reactions, extracts were heat-treated at 85 °C for 10 min (data not shown). The arrow indicates the internal control for PCR efficiency. TSR8 corresponds to the internal quantification standard. (B) Purity of mitochondrial extracts was verified by Western blot analysis probing mitochondrial protein extracts (lanes 2–7, respectively, MRC-5, 184 hTERT, 90P hTERT, MRC-5 hTERT, NHF hTERT and NHF LXIN hTERT) for the presence of mitochondrial HSP70 and PCNA. Total cellular extract (lane 1) was used as a positive control. Arrow indicates mitochondrial HSP70 (upper panel) and PCNA (bottom panel). (C) Gel corresponds to TRAP of mitochondrial extracts containing 3 µg of MRC (negative control, lane 1), 3 µg of MRC-5 hTERT (lane 3), 90P hTERT (lane 5), 1.5 µg of NHF hTERT (lane 7), NHF LXIN hTERT (lane 9) and 184 hTERT (lane 11). Lanes 2, 4, 6, 8, 10 and 12 depict heat-treated extracts and lanes 13–16 the controls of the assay.

To test whether active telomerase could be identified in mitochondrial extracts of our hTERT-overexpressing cells, we performed TRAP assays on total protein extracts from purified mitochondria of the three hTERT-expressing fibroblasts studied here. In addition, two independently derived epithelial cell lines (90P hTERT and 184 hTERT) were included to assess whether cell types other than fibroblasts would also show mitochondrial telomerase activity. Total mitochondrial proteins of these five cell lines were extracted and then submitted to: (i) Western blot analysis to confirm purity (or nuclear contamination) (Fig. 4B) and (ii) TRAP (Fig. 4C). As can be seen in Fig. 4(C), all the mitochondrial samples assayed showed a TRAP product, indicating that telomerase is active in the mitochondria of hTERT-expressing cells. To rule out the possibility that the enzymatic activity observed was due to the presence of gamma polymerase, the only known mitochondrial DNA polymerase, mitochondrial proteins of a cell line that does not express hTERT in the nucleus (MRC-5; Fig. 4A) were included. No PCR product was generated in this cell strain (Fig. 4C, lane 1).

hTERT in the mitochondria sensitizes human cells to H2O2-induced mtDNA damage

The presence of active telomerase in mitochondria suggested a link to our previous observation that NHF hTERT displayed higher levels of mtDNA damage than we have observed previously for other cell lines (Santos et al., 2003; see also Yakes & Van Houten, 1997). To address the question of whether hTERT leads to increased H2O2-induced DNA damage, all isogenic parentals and hTERT cells were exposed to 200 µm H2O2 for 15 and 60 min, and lesions introduced in both mitochondrial and nuclear genomes were evaluated using quantitative polymerase chain reaction (QPCR) (Yakes & Van Houten, 1997; Ayala-Torres et al., 2000; Santos et al., 2002, 2003). H2O2 led to a greater decrease in quantitative amplification of the mtDNA of hTERT-expressing cells than in parental cells (example depicted in Fig. 5A), indicating that higher levels of mtDNA damage are induced after short and long exposures to H2O2 in hTERT cells (Fig. 5B–F). On average, ectopic hTERT expression led to two- to three-fold more mtDNA lesions. For example, at 60 min the NHF hTERT cell line suffered ∼4 lesions/10 kb – two times more mtDNA lesions than the parental NHF cells. Interestingly, no nDNA damage was observed in any of the cell lines studied using gene-specific QPCR. Differences in proliferative characteristics could possibly account for these results. However, β-galactosidase staining demonstrated no significant differences in the percentage of senescent cells between primary and immortalized cell lines. In addition, BrdU incorporation index demonstrated that the majority of cells are cycling (data not shown). Thus, among the five hTERT-expressing cell lines, telomerase activity leads to increased mtDNA damage as compared with the isogenic parental strains. Furthermore, the overall increase in damage appears to plateau at higher levels of total telomerase activity (Fig. 5G).

Figure 5.

Mitochondrial DNA damage is greater in hTERT cells. (A) Relative amplification of the mtDNA of NHF and NHF hTERT cells exposed to 200 µm of H2O2 for 15 min or 60 min was used to estimate the lesion frequency (per 10 kb of the genome) of parental and hTERT counterpart (see Experimental procedures), which is represented in B. C–F show the lesion frequency in the other pairs of cells studied. Student's t-test was performed by comparing the number of lesions in the parental strains to their respective telomerase-expressing derivatives. (G) Graph represents the lesion frequency estimated in the mtDNA of hTERT-expressing cells after a 60-min exposure to 200 µm of H2O2 as a function of telomerase activity as gauged by TRAP. Telomerase activity is expressed as TGP (telomerase generated product) and corresponds to the number of template extended by the enzyme with at least four telomeric repeats (see Experimental procedures). Vertical error bars refer to the lesion frequency and horizontal error bars refer to TGP.

Transfections can result in very high, non-physiological levels of protein expression. To test whether the effects of hTERT on mtDNA sensitivity can be observed when the enzyme is expressed at biological levels, we submitted two epithelial cell lines that express endogenous hTERT to the same H2O2 treatments described above. Cell lines B5 and 184 E6E7 were derived from the 184 strain by treatment with benzo[A]pyrene and viral transformation, respectively. Presence of hTERT transcripts was confirmed by RT-PCR (data not shown) and enzyme activity verified by TRAP. Figure 6(A) shows the weak levels of telomerase activity in B5 (lane 1) and 184 E6E7 cells (lane 3) as compared with the overexpressing isogenic counterpart 184 hTERT (Fig. 6B, lane 1). Nonetheless, these levels of hTERT are sufficient to affect the mtDNA because the amount of mtDNA damage in both B5 and 184 E6E7 cells is similar to that previously seen with 184 hTERT-transfected cells (Figs 6C and 5E). These data suggest that biologically relevant levels of telomerase affect the mitochondrial genome.

Figure 6.

Low telomerase activity affects mitochondrial DNA. Telomerase activity was assayed on 1.5 µg of total cellular proteins from (A) B5 and 184 E6E7 cells, and (B) 184 hTERT. (C) Mitochondrial DNA damage was assayed in B5 (dashed bars) and 184 E6E7 cells (gray bars) after H2O2 treatments using QPCR. Dashed lines correspond to lesions induced in the parent 184 strain submitted to the same type of treatment. Results depict the mean of two biological experiments with corresponding error bars.

Because the detection limit of the QPCR assay is around one lesion per 105 bases, we followed nDNA damage by the phosphorylation status of histone H2AX. The four serine residues from the carboxyl terminus of H2AX become phosphorylated in response to ionizing radiation and other agents that introduce double strand breaks (DSBs) on the DNA. This phosphorylation event takes place in approximately 2 min; it is mediated by the ataxia telangiectasia-mutated pathway and tightly correlated with the formation and presence of DSBs (for review see Redon et al., 2002). As demonstrated in Fig. 7, the parental MRC-5 and NHF as well as their hTERT derivatives show H2AX phosphorylation. Intriguingly, whereas the NHF LXIN hTERT show the strongest signal, no phosphorylation of H2AX was observed in its parental counterpart. These results indicate that DNA damage in the form of DSBs occurs after exposure to 200 µm H2O2 in the nuclear genome of five out of six cell lines examined. Thus, although the mitochondrial genome is more sensitive to telomerase-potentiated H2O2 damage, even when hTERT is expressed at very low levels, telomerase only increased nDNA damage when its expression is high. These data also demonstrate the ability of H2O2 to cause DNA DSBs.

Figure 7.

H2AX phosphorylation as a signal of nDNA damage. Fluorescence signal of phosphorylated H2AX was detected immediately after the H2O2 treatments using an FITC-labeled antibody. The IgG isotype was included to evaluate antibody specificity (data not shown). Signals from control cells are in green; pink and blue lines represent cells treated for 15 and 60 min, respectively.

hTERT expression sensitizes human cells to H2O2-induced cell death

The impact of increased mtDNA damage on cell viability was monitored by following 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) reduction in all fibroblasts, 24 h after the treatments. Prolonged H2O2 exposure (60 min) significantly decreased cell viability of NHF hTERT and NHF LXIN hTERT cells (15 and 30%, respectively) as compared with the isogenic parental strains (see Fig. 8). Interestingly, no significant differences were observed for the parental MRC-5 vs. its hTERT derivative (data not shown), which is the fibroblast cell line showing least telomerase activity as gauged by TRAP (Fig. 4A). These data imply that H2O2-induced mtDNA damage leads to loss of cell viability, and that high expression of hTERT elevates the risk posed to cells by oxidative stress. Taken together with our previous study (Santos et al., 2003), we suggest that this increased mtDNA damage can lead to loss of mitochondrial membrane potential, activation of caspase 3 and apoptosis.

Figure 8.

Decreased cell viability in cells with high hTERT activity. Cell viability was assayed by following MTT reduction in the fibroblasts treated as described earlier and allowed to recover for 24 h. Following this period, medium was aspirated and cells were incubated with MTT for an additional hour. Cells were then lysed and absorbance measured at 570 nm. Statistically significant decreases were observed only for the 60-min treatments and P-values plotted accordingly. NHF (▪), NHF hTERT (□), NHF LXIN (▴) and NHF LXIN hTERT (▵).

Mechanisms of telomerase-mediated sensitization to H2O2

The data described above raise the question of how telomerase can increase the levels of H2O2-induced mtDNA lesions, which ultimately leads to sensitization of the cell to oxidative stress. Our first hypothesis was that the presence of hTERT in mitochondria modulates H2O2 production or breakdown. In fact, inhibition of glutathione pools with 1-chloro-2,4-dinitrobenzene can enhance H2O2-induced mtDNA damage in yeast (Santos et al., 2002). Along the same lines, overexpression of catalase was shown to increase resistance of HL-60 cells to oxidants (Lenehan et al., 1995). Using a fluorescent assay, we identified that all hTERT derivatives decompose H2O2 with the same efficiency as their isogenic parental strains (Fig. 9). Note that H2O2 is freely diffusible and therefore we can expect similar levels of this oxidative stressor throughout the cellular compartments. In addition, no differences in H2O2 generation were detected when comparing telomerase expressors with primary counterparts (data not shown). These results indicate that hTERT does not interfere with the breakdown of H2O2 nor stimulates its production in the cell.

Figure 9.

Efficiency in H2O2-scavenging activity in normal and hTERT-transfected cells. Aliquots of media were removed at indicated times and the amount of H2O2 was measured using Amplex Red®. Fluorescence was read and the concentration of H2O2 was calculated based on a standard curve. The upper graph represents breakdown of H2O2 by the different fibroblasts: (▪) medium alone, (▴) MRC-5, (▵) MRC-5 hTERT, (•) NHF, (○) NHF hTERT, (◆) NHF LXIN and (◊) NHF LXIN hTERT. Bottom graph shows H2O2 decomposition by the epithelial lines: (▪) HANK's alone, (•) 90P, (○) 90P hTERT, (▾) 184 and (▿) 184 hTERT.

Our second hypothesis was that cells expressing hTERT have more bioavailable intracellular divalent metal ions such as iron or copper. This hypothesis is based on the fact that the detrimental effects of H2O2 are believed to result mainly through Fenton chemistry (Fridovich, 1998), in which H2O2 reacts with reduced iron (or copper) giving rise to the potent hydroxyl radical. Because total iron or copper measurements are not informative of the status (free or bound) of these metals in the cell, we used the amount of desferrioxamine (DFO) – a divalent metal chelator – necessary to abrogate the DNA damaging effects of H2O2 as a measure of bioavailable divalent metal ions. This approach has been used extensively in both in vitro and in vivo studies (for reviews see Halliwell, 1989; Donfrancesco et al., 1996; Tam et al., 2003). Figure 10 illustrates the titration experiments performed in the different pairs of fibroblasts. As shown, greater amounts of DFO are necessary to abolish H2O2-induced mtDNA damage in hTERT-expressing lines as compared with their parental strains. Note in particular that the greatest difference in amount of DFO required (20-fold) occurs between the parental NHF LXIN and its hTERT derivative (Fig. 10C). NHF LXIN hTERT was also the transfected cell line that showed the highest level of TRAP activity (Fig. 4A), and the highest difference in the signal of H2AX phosphorylation (Fig. 7). These results, together with the data obtained from the TRAP assay, suggest that the higher the activity of the enzyme, the higher the levels of chelatable metal ions in the cells (Fig. 10D). We were unable to perform this kind of experiment with the epithelial strains, possibly due to their requirement for iron to grow in culture. To ascertain that the effects observed in hTERT-expressing cells could be produced by elevated available iron, parental strains were loaded with iron citrate (based on the amount of metal chelated in the respective hTERT derivatives) prior to H2O2 treatments. These experiments revealed that increasing the iron load of the parental cells to that inferred in the hTERT counterparts led to equal amounts of mtDNA damage (data not shown), supporting the hypothesis that hTERT expression alters H2O2-mediated mtDNA susceptibility by changing divalent metal ion homeostasis.

Figure 10.

Increased levels of chelatable iron are found in hTERT cells. Titration of DFO concentration needed to abrogate the H2O2 effects in the various pairs of fibroblasts (A, B and C). Cells were treated with different concentrations of DFO 24 h prior to the H2O2 exposures. Immediately after the treatments, cells were harvested and DNA was analysed by QPCR. Results depict lesion frequency per 10 kb as a function of DFO, and were based on the levels of DFO used to restore amplification of the 8.9-kb mitochondrial fragment after a 60-min challenge with H2O2. Panel D represents telomerase activity (by TRAP) as a function of the difference in DFO amount used to chelate iron in parental and hTERT derivatives. Standard errors refer to TGP.


To date, the majority of biological properties ascribed to hTERT have been limited to its effects in the nuclear genome. In this study, we present compelling evidence that hTERT is targeted to the mitochondria of human cells through the N-terminal 20 amino acid residues (Figs 1–3), resulting in active telomerase in this organelle. Furthermore, the presence of hTERT in several human cell lines sensitized those cells to mtDNA damage caused by H2O2 exposure. Finally, we present evidence suggesting that hTERT potentiates the damaging effects of H2O2 on mtDNA at least in part via modulation of metal homeostasis.

The mitochondrial localization of hTERT had not been described previously. This could reflect the fact that most of the green fluorescent protein (GFP) constructs made with the catalytic component of telomerase have the GFP tagged to the N-terminus of hTERT, which corrupts the mitochondrial leader sequence and therefore abrogates localization to the mitochondria. Although no reports have clearly demonstrated hTERT localizing in the mitochondria, some recent studies that used hTERT–GFP fusions observed that hTERT is present, at various levels, in the cytoplasm (Ambruster et al., 2001; Haendeler et al., 2003, 2004). Another point to consider is that telomerase has been extensively studied, particularly in ciliates and yeast – organisms in which the TERT gene apparently lacks a mitochondrial localization signal (Fig. 1). Alternatively, one might also explain this mitochondrial localization as being an artefact of high levels of hTERT expression. However, our data obtained with the B5 and 184 E6E7 cells (Fig. 6) argue against this possibility.

Our results clearly point out that hTERT expression increases the damaging effects of H2O2. Greater levels of mtDNA damage were consistently found in hTERT-overexpressing cells exposed to H2O2 as compared with their normal counterparts (Fig. 5A–F) and in cells that have telomerase spontaneously re-activated (Fig. 6), as one might expect in the context of a tumour. The kinetics of H2O2 decomposition (Fig. 9) indicated that expression of hTERT does not modulate H2O2-scavenging activity in mammalian cells, nor does it increase the basal steady-state amounts of H2O2 (data not shown). However, it is possible that hTERT could interfere directly with mtDNA repair or even participate in this process.

Telomerase-expressing fibroblasts required more DFO to abrogate H2O2-mediated mtDNA damage (Fig. 10) than did their non-telomerase-expressing isogenic counterparts, an indication that hTERT cells have more divalent metal ions available to drive Fenton chemistry. Because the stability constant for formation of the Fe(III) complex with DFO is 1031– tens of orders of magnitude greater than that for Cu2+ (1014) or Zn2+ (1011) (Halliwell, 1989), we believe that hTERT is modulating iron levels in hTERT-expressing cells. Levels of iron in mammalian cells are tightly regulated. Most of the iron is associated with enzymes or specialized transport and storage proteins, and it is expected that few, if any, of these metal ions are free in the cell (Aisen et al., 1999). Nonetheless, a distinct fraction of iron is bound to a heterogeneous population of organic anions, polyfunctional ligands and surface components of membranes or extracellular matrix. Collectively, this forms the labile iron pool (LIP), a source of iron that is readily available for incorporation into proteins, which is also involved in free radical formation (Esposito et al., 2002). Taking into account that we did not see modulation of iron-regulatory genes by RT-PCR and flow cytometry analysis (data not shown), we judge that hTERT is not interfering with overall cellular iron uptake or storage, but is rather affecting the LIP.

The relationship between degree of telomerase activity and levels of iron chelated in hTERT-expressing cells (Fig. 10D) is interesting. It has been suggested that telomere-like sequences might serve as an iron sink to protect the rest of the genome from oxidative stress (Henle et al., 1999). Assuming that hTERT-expressing cells maintain longer telomeres, we could speculate that the hTERT cells examined in this study have elevated LIP due to the maintenance of telomere length. It is possible that the H2AX phosphorylation observed (Fig. 7) may be due to DSBs in the telomeres. This might explain the lack of detection of nDNA lesions near the β-globin gene using the QPCR assay. The frequency of DNA lesions calculated with this assay is based on a Poisson distribution, which assumes that damage is randomly distributed along the genome. Alternatively, telomerase itself could require iron for its activity. It is not known whether iron can induce hTERT. However, the amino acid sequence of the hTERT gene reveals the presence of 29 cysteines, residues known to bind iron. Mitochondria are tightly involved in iron trafficking in the cell. In addition to heme synthesis, an increasing body of evidence suggests that Fe–S clusters are synthesized only inside this organelle and are locally used by mitochondrial enzymes or exported for insertion in cytosolic and nuclear enzymes (Arosio & Levi, 2002). Thus it is feasible that mitochondrial hTERT could modulate iron levels within the mitochondria.

One possible biological effect of a higher degree of mtDNA damage in hTERT-expressing lines is decreased cell viability (Fig. 8). Haendeler et al. (2003) recently showed that H2O2 triggers nuclear export of hTERT, amplifying cellular sensitivity of 293 cells to reactive oxygen species (ROS)-induced apoptosis irrespective of the effects on telomere length reduction and cellular senescence. Thus, it appears that movement of hTERT out of the nucleus and, judging by our results, into mitochondria, may play a role in apoptosis. Taking our data into consideration, hTERT may be implicated in apoptosis by increasing the oxidative burden in the mitochondria, resulting in mtDNA damage. In support of this idea, BJ (human foreskin) hTERT fibroblasts were more sensitive than the parental BJ cells to apoptosis (mediated by mitochondrial membrane depolarization) induced by Cr(VI), a known genotoxin that also leads to oxidative stress (Pritchard et al., 2001).

The present observation that hTERT is recruited into the mitochondria has broad implications. First, it was shown that telomerase expression in mice with ample telomere reserve renders the animals more susceptible to papilloma formation when treated with chemical carcinogens (Gonzalez-Suarez et al., 2001). Robust telomerase activity was also found to be associated with spontaneous development of mammary intra-epithelial neoplasia and invasive mammary carcinomas in aged female mice (Artandi et al., 2003). These authors suggest that the activity of telomerase during tumorigenesis may serve both for stabilization of telomeres and another function that does not depend on telomere reserve. Our findings might help understand these data by showing the additional feature of telomerase: an increased sensitivity to oxidative stress mediated through the mitochondria. Mitochondrial DNA damage can have various effects in the cell, including an amplification of the steady-state level of ROS (Yakes & Van Houten, 1997; Santos et al., 2003) as well as a higher rate of mtDNA mutations. In this context, it is interesting to note that many telomerase-positive human cancers have mtDNA mutations (for a recent review see Copeland et al., 2002).

Second, the only DNA polymerase so far detected in mitochondria is gamma polymerase. This enzyme is known to function both in mtDNA replication and in mtDNA repair. Could TERT play a role in mtDNA transactions? Interestingly, the well-established strand-displacement model of human mtDNA replication has recently been re-examined. This model posits that leading-strand replication begins at closely spaced, defined sites and proceeds unidirectionally with displacement of the parental leading strand until approximately two-thirds of the closed circular mtDNA has been copied. As a consequence, the replication fork passes a major origin for lagging-strand synthesis, exposing it in a single-stranded form (for a review see Clayton, 2003). However, recent data from two-dimensional gels suggest that replication of mammalian mtDNA is strand-coupled. There is also speculation that replication for mitochondrial genome maintenance differs from that of genome repopulation (Yang et al., 2002). Telomerase was demonstrated to have error-prone polymerase activity, with a misincorporation rate of 2 × 10−3 per nucleotide (Kreiter et al., 1995). In addition, telomerases were shown to bind and elongate almost any G-rich single-stranded DNA containing a free 3′ end in vitro (Greider, 1996). There are 98 G triplet clusters in the human mitochondrial genome. Intriguingly, whereas the light strand of the human mtDNA has one TTAGGG telomeric-repeat-like sequence, the heavy strand has 21 of these repeats, with one of them being in the D-loop – the region where mtDNA replication begins.

Mitochondrial telomerase may play a role in the mitochondrial base excision repair machinery, working on transient nicks created by this system to repair oxidative lesions. Alternatively, TERT could work on the capping of H2O2-induced DSBs in the mitochondrial genome, analogous to what occurs in the nuclear DNA. In the mitochondria, however, the persistence of such breaks could lead, ultimately, to cell death. Interestingly, TRF-2, one of the proteins involved in telomere capping, also has a predicted mitochondrial leader sequence (J.H.S. and B.V.H., unpublished observation). In addition, a major consideration as far as telomerase activity in mitochondria is concerned relates to whether its RNA component is also present in this organelle. Our results from the TRAP assay suggest that hTR is also transported into the mitochondria.

Finally, the observation that only the TERT gene of higher eukaryotes has the mitochondrial-targeting sequence points to some specialized function of telomerase in the organelle of more complex organisms.

Experimental procedures

Cells and cell culture

The hTERT cell lines derived from the parental strains NHF, MRC-5, 90P and 184 were constructed in the Laboratory of Molecular Carcinogenesis at the NIEHS. Briefly, the retroviral vector pLXIN (provided by A. D. Miller, Fred Hutchinson Cancer Center) containing hTERT was transfected into the ecotropic packaging cell line GPE86. Supernatant generated was then used to infect the amphotropic packaging cell line PT67 in the presence of 4 µg mL−1 polybrene. PT67-infected cells were grown in selection medium for 2 weeks and pooled for supernatant collection. NHF, MRC-5 cells (ATCC) and normal human breast epithelial strains 184 and 90P (provided by Martha Stampfer, Lawrence Berkeley National Laboratory) were infected with pLXIN–hTERT virus and selected in geneticin-G418 (800 µg mL−1). Resistant colonies were pooled and tested for telomerase activity as well as immortalization (greater than 100 population doublings). Characterization of the epithelial strains 90P and 184 can be found in Annab et al. (2000). Epithelial cells 184 E6E7 and B5 were derived from 184 cells by viral infection and benzo[a]pyrene treatment, respectively (Martha Stampfer). NHF LXIN cells and their hTERT counterpart were a gift from Dr William K. Kaufmann (UNC, Chapel Hill) and were derived as described elsewhere (Heffernan et al., 2002). Note that the parental cells carry the empty vector. All fibroblasts were routinely grown and subcultured as described previously (Santos et al., 2003). Epithelial cells were grown and subcultured as explained in in Annab et al. (2000). Primary strains used for experiments were kept in early passage (no more than 25 population doublings). HeLa cells (ATCC) were maintained in MEM (Gibco) supplemented with 10% FBS and 2 mm glutamine.

Generation of EGFP fusion proteins and cell transfections

The full-length hTERT gene was PCR amplified from the vector LXIN–hTERT (a gift from Dr J. Carl Barrett) used to infect the cells described above. The gene was cloned into the N-terminus of the pEGFP-N1 vector (Clontech) by restriction enzyme digestion. After cloning, the fusion vector was sequenced, and used to infect primary NHF cells (that do not express endogenous hTERT) and HeLa cells, which express endogenous hTERT. Cells were transiently transfected with the fused hTERT–EGFP plasmid using Lipofectamine 2000 (Invitrogen), as suggested by the manufacturer, onto 60-mm dishes equipped with a coverslip attached to the bottom (MatTek). Proficiency of the method ranged from 1 to 5% for the primary cells and 60 to 80% for HeLa. The first 20 amino acids of hTERT were added by PCR to the N-terminus of EGFP, and the fused protein subcloned into the pCMV-myc and pCMV-HA vectors (Clontech). Constructs were sequenced and transiently transfected in HeLa cells as above described.

Confocal microscopy

Subcellular localization was evaluated 24 h after transfections. Immediately before visualization, cells were incubated for 5 min with 5 nm of Mitotracker red (Molecular Probes) to stain mitochondria, and washed three times with PBS prior to analysis. A laser scanning confocal microscope (LSM 510 mounted on an Axiovert 200M microscope, Carl Zeiss) was used to obtain fluorescence images. The objective lens used was the PlanApo 63×/1.4 oil-immersion DIC, and the pinhole was set to achieve a z-resolution of 1.0 µm.

Isolation of mitochondrial proteins

Mitochondrial proteins were obtained as described earlier (Graziewicz et al., 2002). Briefly, cells were harvested, washed with PBS and resuspended in HDB buffer (5 mm KPO4, pH 7.5, 2 mm MgCl2, 1 mm 2-mercaptoethanol). Cells were left on ice for 60 min to swell and broken by dounce homogenization. Sucrose buffer (0.7 m sucrose, 12.5 mm Tris (pH 7.5), 12.5 mm EDTA, 12.5 mm MgCl2) was added and cells centrifuged for 5 min at 500 g to remove nuclear contaminants. Supernatants were submitted to further centrifugation (45 min at 27 000 g) to collect mitochondria. Mitochondrial pellets were resuspended in 1 mL HDB buffer and submitted to sucrose gradient separation as described by Bogenhagen & Clayton (1974). The mitochondrial fraction was resuspended in lysis buffer (1% Triton X-100, 0.3 m NaCl, 10% glycerol, 20 mm Tris/Cl (pH 8.0), 14 mm 2-mercaptoethanol) and lysates were centrifuged for 2 h at 23 000 g. Immediately after centrifugation, supernatants were collected and protein concentration estimated. Purity of these extracts was verified by Western blots. These mitochondrial lysates were kept at −80 °C and were used for determination of telomerase activity in mitochondria.

Western blot analysis

The mitochondrial protein extracts (10 µg) described above were probed for mitochondrial enrichment using an antibody for the mitochondrial heat-shock protein 70 (HSP70, Affinity Bioreagents) and analysed for nuclear protein contamination with an antibody anti-PCNA (Santa Cruz Biotechnology). A sample of crude extract (5 µg) was included as a positive control.

Telomerase activity (TRAP)

Total cellular and mitochondrial protein extracts were assayed for telomerase activity using a PCR-based protocol as described previously (Kim et al., 1994) with some modifications (TRAPeze, Chemicon). Telomerase activity was calculated as indicated by the manufacturer and is expressed as TGP (telomerase generated product), corresponding to the number of template extended by the enzyme with at least four telomeric repeats.

H2O2 and desferrioxamine treatments

H2O2 (Sigma) treatments were performed as described by Santos et al. (2003). Briefly, cells were challenged with 200 µm H2O2 for either 15 or 60 min, and harvested immediately after treatments for DNA analysis. Because H2O2 breaks down rapidly in MEBM medium (data not shown), all H2O2 exposures of epithelial cells were carried out in Hank's salt solution. When desferrioxamine (DFO; Sigma) treatments were conducted, cells were maintained for 24 h in medium containing all the supplements and different concentrations of DFO prior to H2O2 exposure.

DNA isolation and DNA damage analysis

High-molecular-weight DNA was extracted and QPCR performed and analysed as described elsewhere (Ayala-Torres et al., 2000; Santos et al., 2002). Briefly, genomic DNA is isolated and specific primers are used to amplify a fragment of the mitochondria and/or nDNA. The assay is based on the premise that DNA lesions block/slow the progression of the polymerase so that only undamaged templates can participate in the PCR reaction. Thus, amplification is inversely proportional to DNA damage; the more lesions on the target DNA the less amplification. Amplification of treated samples is then compared with controls and relative amplifications calculated. These values are then used to estimate the average number of lesions per 10 kb of the genome, assuming a Poisson distribution (Ayala-Torres et al., 2000; Santos et al., 2002). The sequence of the primers used in the present study can be found elsewhere (Santos et al., 2002). Note that large (8.9 kb) and small (139 bp) fragments of the mtDNA are amplified. The latter is used to monitor the copy number of the mitochondrial genome and also to normalize the results obtained with the large fragment (for more details see Santos et al., 2002). Results presented here are the mean of, at least, two biological experiments. Student's t-test was performed to evaluate statistical significance.

Histone H2AX assay

Fibroblasts were treated as described above and immediately collected for H2AX phosphorylation analysis, using the flow cytometry kit available from Upstate. The protocol followed was as recommended by the manufacturer and cells were analysed using a Becton Dickinson FACSort.

Viability assessment

All fibroblasts were assayed for viability by monitoring the reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT, Sigma), which occurs primarily by mitochondrial succinate dehydrogenase, as described in Yakes & Van Houten (1997). Results represent the mean of at least two biological experiments.

H2O2 depletion from cultured media

H2O2 detection in the media was performed using the Amplex Red® Kit (Molecular Probes). The assay is based on the detection of H2O2 using 10-acetyl-7-hydroxiphenoxazine (Amplex Red reagent) a highly stable and specific probe for this compound. For our purposes, aliquots of media were taken at 0, 5, 15, 30 and 60 min after addition of H2O2 and the assay was performed following the manufacturer's instructions. Controls with catalase were also included to ensure that the fluorescence signal was due to H2O2.


We would like to thank Ms Melissa J. Humphrey for assistance with DNA extractions and quantifications, and the members of the Van Houten and Copeland laboratories for fruitful discussions. We would also like to thank Mr Jeffrey M. Reece for assistance with the confocal images, Dr William K. Kaufmann (UNC, Chapel Hill) for providing the cells NHF LXIN and its hTERT derivative, Dr Hidetoshi Tahara (Hiroshima University, Japan) for generating the other hTERT fibroblasts used, Dr Martha Stampfer (Lawrence Berkeley National Laboratory) for the B5 cells, and Drs Matthew J. Longley, William C. Copeland and Thomas A. Kunkel for critical review of the manuscript.