Aging Cell

Disruption of the NES of hTERT abolishes its ability to translocate out of the nucleus

Nuclear export of hTERT is mediated by 14‐3‐3 and the nuclear exportin CRM1 (Seimiya et al., 2000). In trying to create a mutant that does not translocate into mitochondria, we disrupted the NES of hTERT by site‐directed mutagenesis. A putative NES had been previously identified on hTERT but the exact amino acid residues required for nuclear export of the protein were not identified (Seimiya et al., 2000). The software package NetNES1.1 predicted that leucine (L) substitutions at positions 980 and 987 would suffice to disrupt the NES of hTERT. We replaced these residues with alanine (L980A/L987AhTERT) and established the subcellular localization of the mutant protein using confocal microscopy, and Western blots in subcellular fractionations.

Mutant and wild‐type (WT) proteins were fused to enhanced green fluorescent protein (EGFP) and transiently transfected into HeLa cells. L980A/L987AhTERT‐EGFP was, as expected, nuclear and pretreatment for 2 h with 20 nm of leptomycin B, a potent inhibitor of the nuclear export machinery, did not impact its subcellular localization (Fig. 1A lower panels). Conversely and in agreement with previous work (Santos et al., 2004, 2006a; Ahmed et al., 2008), WT hTERT was both nuclear and cytoplasmic, including mitochondrial, but upon incubation with leptomycin B it lost its cytoplasmic localization in most cells (Fig. 1A upper panels). About 80% of cells expressing the WT protein showed hTERT in both nucleus and cytoplasm (including mitochondria) whereas in the mutant‐expressing cells 90% of hTERT was nuclear. Subcellular fractionations in cells stably expressing the proteins showed that the amount of hTERT in the mitochondria of the mutant was negligible (25‐fold lower than in cells expressing the WT protein; Fig. 1B). Herein, this mutant is referred to as _NES‐hTERT.

Expression of _NES‐hTERT causes premature senescence in normal diploid fibroblasts

Having confirmed that _NES‐hTERT was nuclear, we stably expressed it in primary human foreskin fibroblasts (NHF) that are telomerase‐negative, using retroviral transduction. We selected these cells because we had previously characterized WT hTERT in this cellular background (Santos et al., 2004, 2006a). Unexpectedly, we found that NHF‐expressing _NES‐hTERT arrested growth prematurely, that is earlier than non‐hTERT parental irrespective of the vector in which it was expressed; doubling time of the cells increased significantly (Fig. 2A left panel). The same was observed when the mutant was expressed in another background strain GM7532 (from Coriell Repository; Fig. 2A right panel). Growth arrest was accompanied by flattened and enlarged morphology reminiscent of senescent cells (Fig. 2B). DNA profiles using propidium iodide (PI) and flow cytometry indicated that the cells were arrested in the G1/S transition (Fig. 2C left panel), which is typical of the growth arrest associated with senescence (Campisi & d’Adda di Fagagna, 2007). Analysis of the senescence markers senescence‐associated β galactosidase activity (SA β‐gal, Fig. 2B), p21^WAF−1 (or p21) and p16^INK4a (referred herein as p16, Fig. 2D) confirmed that NHF‐expressing _NES‐hTERT were senescent. Single‐cell analysis showed that virtually all cells were positive for p21, and counterstaining of the same cells for p16 revealed that most but not all cells that were positive for p21 also upregulated p16 (example depicted in Fig. 2D). p21 upregulation was confirmed by immunoblots (Fig. 2E). We also detected polynuclei in cells expressing the mutant protein (about 5–10% of the cells, Fig. 2F). In all cases, cells shared the same enlarged cytoplasm. Multinucleated cells were not observed in non‐hTERT or WT‐expressing controls (data not shown).

Upregulation of p21 suggested the involvement of p53, a protein that participates in the growth arrest of senescent cells (Campisi & d’Adda di Fagagna, 2007). We therefore tested whether p53 indeed played a role in the cell cycle blockade triggered by expression of _NES‐hTERT by expressing empty vector, WT and mutant hTERT in GM847, an SV40‐immortalized cell line in which both the p53/p21 and the p16/Rb senescence pathways are disrupted. In these cells, no cell cycle blockade was observed (Fig. 2G), suggesting a role for p53 in arresting _NES‐hTERT‐expressing cells. GM847 cells carrying _NES‐hTERT were maintained for continuous growth for two months, and no changes in phenotype or in the rate of cell division were observed (data not shown).

These results were intriguing given that _NES‐hTERT was localized to the ‘correct’ subcellular compartment. Under our culture conditions, NHF undergo replicative senescence at about population doubling (PD) 50–60; most cells were infected at around PD24. The senescent phenotype observed in _NES‐hTERT‐expressing cells was usually observed earlier than PD40 and irrespective of the PD in which the cells were infected. In some independent infections, _NES‐hTERT‐expressing cells senesced as early as at PD28 (see example on Fig. 2A left panel). The same premature growth arrest was observed in a different strain (GM7532) although in this cellular background the growth defect was consistently observed later (see example on Fig. 2A right panel). GM7532 senesce between PDs 60 and 70 in our culture conditions (data not shown). As no additional exogenous stress was imposed upon the cells, these observations suggest that the translocation of hTERT between subcellular compartments is required for cellular immortalization.

It is possible that mutations in _NES‐hTERT disrupted the protein expression levels or catalytic activity, ultimately leading to telomerase enzymatic activity below the threshold required for cellular immortalization. To test this, we analyzed the levels of hTERT mRNA and telomerase enzymatic activity in whole cellular extracts using, respectively, RT‐PCR and the telomeric repeat amplification protocol (TRAP). As shown in Fig. 3A, not only the RNA of the mutant protein was expressed at the same levels as WT hTERT but also _NES‐hTERT was catalytically active as judged by the TRAP.

Various manipulations on hTERT were shown to lead to biological inactivity (i.e, inability to extend telomeres in vivo) despite enzymatic activity in vitro (Counter et al., 1998; Armbruster et al., 2001; Banik et al., 2002). Thus, it is possible that _NES‐hTERT while active as gauged by the TRAP is biologically inactive. To test this hypothesis, we measured mean telomeric length by Southern blotting based on the mean telomere restriction fragment technique (mTRF). Cells were freshly derived at PD15 and allowed to undergo 8 PDs prior to telomere measurements; at PD23, DNA was isolated from parental and the hTERT derivatives. Results are representative of duplicate experiments. Mean telomeric length in parental cells was 6.38 kb (Fig. 3B lane 1) and significantly longer in cells carrying the WT protein (8.66 kb; Fig. 3B lane 2), demonstrating that WT telomerase can efficiently extend telomeres. However, the mean telomeric length in cells carrying the mutant _NES‐hTERT (6.21 kb) was comparable to that of the parental strain (Fig. 3B lane 3), indicating that the mutant is unable to extend telomeres in vivo. These data also show that the expression of _NES‐hTERT per se does not cause telomere shortening.

It had been shown previously that high levels of telomerase activity in apparently nonstressed cells induce premature cellular senescence (Gorbunova et al., 2003). However, in our experiments, enzymatic activity of WT hTERT, which promoted cell growth, and _NES‐hTERT, which promoted senescence, were very similar (Fig. 3A). In addition, protein levels of hTERT in the nucleus of the mutant cells were lower than in the WT (Fig. 1B), making it unlikely that the observed growth inhibitory effects of _NES‐hTERT were attributable to the elevated levels of telomerase. In addition, average telomere length in _NES‐hTERT was comparable to the nonimmortalized control (Fig. 3B) in contrast to the results of Gorbunova et al. (2003) who reported that their hTERT‐positive senescent cells had long telomeres.

Nuclear DNA damage and activation of the DDR are observed only in cells expressing _NES‐hTERT

Mean telomere length was comparable between parental and _NES‐hTERT‐expressing cells as shown in Fig. 3B. Despite this, the mutant cells consistently stopped cycling before the non‐hTERT counterparts, showing that the rapid rate of senescence in NHF _NES‐hTERT was not because of the replication‐based telomere shortening. These observations led to the hypothesis that the telomeres of _NES‐hTERT cells were damaged.

Dysfunctional telomeres, whether caused by gradual telomere erosion because of continuous cell proliferation or by stochastic telomeric DNA damage, induce a DNA double‐strand break (DSB)‐like response with activation of DNA repair proteins such as meiotic recombination 11 (MRE11), Nijmegen breakage syndrome 1 (Nbs1), 53‐binding protein 1 (53BP1), and phosphorylated forms of the ataxia telangiectasia mutated (ATM S1981), checkpoint protein 2 (Chk2 at T68) and histone H2AX (S139). These DNA damage factors form discrete foci at dysfunctional telomeres, which can be visualized by immunofluorescence (IF) (d’Adda di Fagagna et al., 2003). Telomere dysfunction‐induced foci (TIF) are distinguishable from DNA damage foci that form at nontelomeric sites because of the presence of telomeric sequences along with the DNA damage factors (Takai et al., 2003).

To test whether the telomeres of _NES‐hTERT cells were damaged, we probed activation of proteins involved in the DSB‐like response or DDR and quantified TIF in single cells using microscopy. We also evaluated genomic DNA damage as a surrogate of overall chromosome integrity using gene‐specific quantitative PCR (QPCR, Santos et al., 2002, 2006b; Kovalenko & Santos, 2009). The DDR proteins studied included serine 139 phosphorylated H2AX (γH2AX), activated ATM (ATM‐S1981) and Chk2 (N‐T68), as well as phosphorylation of serine 15 (S15) on p53. Controls included NHF (hTERT‐negative) at PD26 and NHF‐expressing WT hTERT.

While very few control or WT hTERT‐expressing cells displayed visible γH2AX foci (fewer than 12%), almost all _NES‐hTERT‐expressing cells displayed at least one visible γH2AX focus (Fig. 4A; quantification on the graph on panel B). γH2AX colocalized with active ATM and Chk2 kinases, and all γH2AX‐positive cells displayed elevated levels of p53S15 (Fig. 4A), demonstrating activation of the G1 DNA double‐strand break checkpoint (Campisi & d’Adda di Fagagna, 2007). None of these markers were detected at significant levels in the non‐hTERT or WT‐expressing control cultures (Fig. 4B and data not shown).

Next, we determined whether DNA damage foci colocalized with telomeres, as performed by us previously (Herbig et al., 2004). A cell was scored TIF positive when 50% or greater of its DNA damage foci colocalized with telomeric DNA sequences (see representative image on Fig. 4C). Consistent with our previous observations (Herbig et al., 2004), about only 10% of early passage NHF was TIF positive (Fig. 4D). Expression of WT hTERT reduced the levels of TIF‐positive cells to < 10%, demonstrating that telomerase efficiently prevents telomere dysfunction in these cells. In contrast, over 60% of _NES‐hTERT‐expressing cells were TIF positive (Fig. 4C and D). Together, the results presented in Fig. 4A‐D show that telomeres in _NES‐hTERT cells trigger a DSB‐like response.

Finally, we estimated the basal levels of nuclear DNA lesions using QPCR as a surrogate of genomic DNA integrity. QPCR has been used extensively by us and others to identify damage in both nuclear and mitochondrial genomes, and has proven particularly useful in the identification of oxidative stress‐mediated damage (Sobol et al., 2002; Santos et al., 2003; Sawyer et al., 2003; Shukla et al., 2003, Sorolla et al., 2008). The QPCR assay measures integrity of DNA using long PCR targets. The results obtained after PCR amplifications of the target DNA are used to estimate the lesion frequency present on each DNA strand based on a Poisson distribution. Presence of lesions reflect that the sample of interest amplifies less than its control while negative number of lesions can be observed when the DNA of a given sample amplifies better than its respective control (for more details on the assay see Santos et al., 2002, 2006b; Kovalenko & Santos, 2009). The nuclear fragment amplified in this analysis was 13.5 kb in length. Assuming that damage is randomly distributed, QPCR allows an overall picture of the integrity of the genome under study. Sensitivity limit of the technique is 1 lesion in every 100 000 bases (Santos et al., 2002, 2006b; Kovalenko & Santos, 2009).

We analyzed nuclear DNA (nDNA) integrity in parental cells and in the WT or mutant derivatives. Experiments were reproduced using independently derived cultures in PDs where no signs of cellular senescence were apparent in either parental or _NES‐hTERT cells. As presented in Fig. 4E, cells expressing _NES‐hTERT had significantly higher levels of nDNA damage than that in the non‐hTERT and the WT‐expressing counterparts, suggesting that telomeres in _NES‐hTERT‐expressing cells likely have DNA lesions.

Impaired mitochondria are found in _NES‐hTERT‐expressing cells

The next obvious question that arose was how could a mutant hTERT that is impaired in subcellular shuttling be involved with nDNA damage in telomeric and nontelomeric sites? We and others have shown that hTERT is present not only in the nucleus but also in mitochondria (Santos et al., 2004, 2006a; Ahmed et al., 2008; Haendeler et al., 2009) and our data revealed almost complete absence of hTERT in mitochondria of the mutant‐expressing cells (Fig. 1). Mitochondria are the main source of endogenous ROS and it has been shown that cells expressing WT hTERT have improved mitochondrial function under normal conditions, including decreased ROS generation (Ahmed et al., 2008; Haendeler et al., 2009). Thus, another potential effect of impaired hTERT shuttling is improper mitochondrial ‘regulation’ and mitochondrial dysfunction.

The mitochondrial genome is a critical target for oxidative damage primarily because of its proximity to the site of ROS generation. Once damaged, mtDNA amplify oxidative stress because of the decreased expression of critical protein components of the electron transport chain. Such effects lead to a vicious cycle of increasing ROS generation and organellar deregulation that can eventually trigger apoptosis (Van Houten et al., 2006; Wallace, 2008). Thus, mtDNA integrity is a good indicator of endogenous oxidative stress and proper mitochondrial function. To test whether mitochondria were impaired in _NES‐hTERT cells, we evaluated different markers of mitochondrial function such as integrity of the mtDNA and basal rates of ROS production. We concomitantly monitored levels of oxygen consumption because it can influence the rate of ROS formation (Murphy, 2009). Finally, we also evaluated the mitochondrial ultrastructure by electron microscopy (EM).

We estimated the basal levels of mtDNA damage using QPCR. One of the advantages of using QPCR to estimate DNA damage is that both nuclear and mitochondrial genomes can be evaluated in the same sample (Kovalenko & Santos, 2009). We thus took the samples used in Fig. 4E, including cell lines derived at independent times in the both NHF and GM7532 backgrounds, to evaluate the integrity of the mtDNA from parental, WT and _NES‐hTERT cells. We added to this analysis samples from freshly derived cells in which we included a senescent control. Senescent cells were shown to carry dysfunctional mitochondria (Passos et al., 2007). At the time of these analyses neither parental nor _NES‐hTERT had signs of cellular senescence (that is, they were still cycling).

As presented in Fig. 5A, mtDNA damage was significantly higher in _NES‐hTERT‐expressing cells when compared to young non‐hTERT and to WT hTERT controls. Levels of mtDNA damage paralleled that of nDNA damage (Fig. 4E). Lesions on _NES‐hTERT cells while high were still statistically lower than their senescent counterparts (Fig. 5A, compare bars on the extreme right; P = 0.05).

We next measured the basal levels of ROS production in the cells using Mitosox, a dye specific for the detection of mitochondrial superoxide anion (O₂⁻). We measured Mitosox fluorescence at a single‐cell level by confocal microscopy and by fluorescence‐activated cell sorting (FACS). As the data were identical with both approaches, we depict results obtained by FACS (Fig. 5B). A negative control in which cells were exposed to 5 μm of the uncoupler FCCP revealed the specificity of the Mitosox signal (data not shown). In agreement with previous reports (Ahmed et al., 2008; Haendeler et al., 2009), lower levels of ROS were detected in cells carrying WT hTERT compared to non‐hTERT controls (Fig. 5B). Differences in oxygen consumption were ruled out as potential explanation for these data (Fig. 5C left panel). No differences were observed in the rate of proton leak between non‐hTERT and its WT derivative (Fig. 5C right panel).

Although senescent cells tend to have higher levels of ROS compared to young culture counterparts, this difference did not reach statistical significance (Fig. 5A, compared bars of extreme left and right). Consistent with higher levels of mtDNA damage in NHF _NES‐hTERT, cells carrying the mutant protein also produced significantly higher basal levels of ROS (Fig. 5B). These results are in agreement with our previous report showing that cells carrying damaged mtDNA produced higher levels of ROS than cells carrying nondamaged mtDNA (Santos et al., 2003). In agreement with dysfunction, the mitochondria of these cells while consuming more oxygen had a high degree of proton leak (Fig. 5C).

The increased ROS detected may explain why cells expressing the mutant undergo premature growth arrest as higher levels of ROS could compound the defect in telomere elongation ultimately accelerating the entrance into senescence. To test this, we cultured parental and _NES‐hTERT‐expressing cells under chronic antioxidant supplementation (NAC, 5 mm) and analyzed the kinetics of cell growth. As shown in Fig. 5D, _NES_hTERT when cultured under NAC continued to proliferate with similar kinetics to the parental controls while, as previously observed (Fig. 2A), they stopped cycling prematurely in the absence of the antioxidant. Of note, _NES‐hTERT was not able to sustain the growth beyond the PD when the non‐hTERT parental senesced even in the presence of NAC (data not shown).

Mitochondrial impairment is often accompanied by changes in organellar morphology such as swelling, loss of cristae and the appearance of megamitochondria (Kim et al., 2007). Corroborating the data mentioned earlier, analysis of mitochondrial ultrastructure by EM showed that the expression of _NES‐hTERT caused marked mitochondrial swelling with loss of cristae, intramitochondrial inclusions and irregular‐shaped organelles throughout the cytoplasm (Fig. 5E and enlarged image at the bottom). No worm‐shaped mitochondria or mitochondria undergoing division (or fission) as observed in NHF and in WT controls were detected in the mutant (Fig. 5E).

Taken together, the data presented in Fig. 5 show that the mitochondria of _NES‐hTERT cells are dysfunctional, supporting a role for hTERT in mitochondrial biology (Santos et al., 2004, 2006a; Ahmed et al., 2008; Haendeler et al., 2009). In addition, they suggest that the telomere elongation defect in the mutant that drives growth arrest is compounded by dysfunctional mitochondrial leading to premature entrance into senescence.

Mitochondrial dysfunction in _NES‐hTERT is not a consequence of cellular senescence

It could be argued that impaired mitochondria observed in cells carrying the mutant hTERT simply reflect the senescent phenotype. We do not favor this hypothesis because the analyses mentioned earlier were performed while cells were still cycling i.e., prior to the appearance of senescence.

To directly test the involvement of senescence to the mitochondrial defects, we analyzed mtDNA damage and ROS production in GM847 fibroblasts retrovirally transduced with empty vector, WT‐ or _NES‐hTERT. GM847 are immortalized cell lines that maintain telomeres through the ALT (alternative lengthening of telomeres) pathway (Stewart et al., 2002). These cells did not stop cycling or undergo senescence as a result of _NES‐hTERT expression given they are p53/p16 impaired (Fig. 2D and data not shown). For mtDNA integrity and ROS production analyses, we used two independently derived cultures expressing the mutant protein. Expression of _NES‐hTERT in GM847 cells also resulted in significant increases in mtDNA damage and ROS production (Fig. 6), indicating that the mitochondrial dysfunction observed in _NES‐hTERT‐expressing cells does not result from the senescence phenotype. As the ALT cells maintain telomeres through a telomerase‐independent pathway, these data support the conclusion that the senescence observed in the primary cells results from the inability of _NES‐hTERT to maintain telomeres.

Residues on the NES regions are required for cellular immortalization

As the only apparent defect of _NES‐hTERT was its inability to translocate out of the nucleus, our data suggest that shuttling of hTERT between subcellular compartments is required for cellular immortalization.

However, it can be argued that the mutations introduced to create _NES‐hTERT affected other properties of the protein, besides its ability to shuttle between nucleus and cytoplasm. For instance, it is possible that the substituted residues caused conformational changes that do not allow hTERT to interact with a protein responsible for localizing it at telomeres and/or bind telomeres (or telomeric DNA) or its associated RNA (hTR). In either case the telomere elongation function of telomerase would be compromised. This scenario, however, is unlikely given that the mutant hTERT was equally efficient as WT hTERT in binding and extending telomeres in vitro as judged by the TRAP (Fig. 3A), and binding sites between hTERT and hTR have been mapped onto the N‐terminus of hTERT (Beattie et al., 2000, 2001; Armbruster et al., 2001; Bachand & Autexier, 2001; Banik et al., 2002).

Structural studies have shown that the C‐terminus of hTERT is involved with catalytic activity and processivity but little was performed specifically in the NES region (Beattie et al., 2000, 2001; Banik et al., 2002; Hossain et al., 2002; Autexier & Lue, 2006; Middleman et al., 2006). While large amino acid substitutions (6 at a time) between amino acids 978 and 992, encompassing the NES, led to mutants unable to immortalize cells because of the complete lack of enzymatic activity, one single amino acid change at position 980 led to a slight defect in enzymatic activity as judged by the conventional assay (Banik et al., 2002, Huard et al., 2003). The latter mutant was not assayed for its ability to elongate telomeres in vivo. Based on these studies, one can conclude that mutations in the NES region are more likely associated with disruption in catalytic activity in vitro and the consequent inability of telomerase to elongate telomeres in vivo. Although our data show that problems with catalytic activity in vitro are not responsible for the effects linked to expression of _NES‐hTERT, it is a formal possibility that amino acids in the NES region are directly involved with the metabolism of chromosomal telomeric substrates by telomerase in vivo besides its role in nuclear‐cytoplasmic shuttling.

The hTERT DAT mutants previously described (Armbruster et al., 2001; Banik et al., 2002) had similar phenotype as reported here for _NES‐hTERT: catalytic activity in vitro but biological inactivity in vivo. However, some differences are noteworthy. For instance, while the size of telomeres were comparable between the parental strains and the _NES‐hTERT or DAT mutants (Fig. 3B, Armbruster et al., 2001; Banik et al., 2002), _NES‐hTERT cells consistently stopped cycling prior to the parental strains (Fig. 2A). Increased ROS generation was reported only in _NES‐hTERT cells (Fig. 5). It is likely that the elevated levels of ROS compounded the telomere defect in this mutant accelerating the entrance into senescence. This conclusion is supported by findings presented in Fig. 5D. Finally, a fraction of cells expressing _NES‐hTERT also presented multiple nuclei and ER stress (Fig. 1F and data not shown) which were not reported previously. Both these phenotypes can result from oxidative stress (D’Angiolella et al., 2007; Hanada et al., 2007) and could be secondary to the mitochondrial dysfunction.

Mutations in the NES region of hTERT are associated with mitochondrial dysfunction

Irrespective of the exact reason why _NES‐hTERT is defective in cellular immortalization, an important finding from these studies was that mutations in the NES of hTERT had a negative impact on mitochondria. We showed that basal levels of mtDNA damage, mitochondrial ROS generation and the rate of proton leak (another marker of mitochondrial dysfunction) were all higher in cells carrying the mutant independent of the cellular background (NHF or GM847) in which it was expressed. The ultrastructure of the mitochondria of _NES‐hTERT cells was also clearly altered (Fig. 5), collectively indicative of mitochondrial dysfunction.

One explanation for these data is that once translated in the cytoplasm, hTERT first translocates into the nucleus to be then exported out to mitochondria. In such case, confinement of hTERT into the nucleus leads to mitochondrial dysfunction simply by the absence of the protein in the organelle. This possibility is consistent with recent results that showed mitochondrial defects in shRNA or knockout cells for TERT (Ahmed et al., 2008; Haendeler et al., 2009). Alternatively, shuttling of hTERT between subcellular compartments may exert an additional level of regulation upon mitochondria. Such effect may be based on increasing the mitochondrial pool (in some or in all cells of the population) or it may involve cross talk and/or translocation with nuclear/cytoplasmic proteins that in turn are required for mitochondrial function.

Another potential explanation for these data is that mutations on the NES region shift the role of hTERT from telomere maintenance to gene expression regulation, possibly affecting mitochondrial‐related genes. In such case, the mutant has gained function. Interestingly, the ability of hTERT to modulate expression of gene(s) based on its RNA‐dependent RNA polymerase activity has been reported recently (Maida et al., 2009).

A gain of function mutation could also explain why _NES‐hTERT‐expressing cells had dysfunctional mitochondria when compared to the parental cells. However, in the latter case one cannot exclude the possibility that _NES‐hTERT exerts some degree of dominant negative activity upon endogenous hTERT in mitochondria. Although most adult somatic cells were thought to be telomerase negative, both hTERT mRNA and protein have been detected in primary strains albeit at low levels (Kilian et al., 1997; Ramakrishnan et al., 1998; Masutomi et al., 2003). Our unpublished results show that the expression of _NES‐hTERT in two telomerase‐positive cancer cell lines also leads to mitochondrial dysfunction, in line with the hypothesis that _NES‐hTERT has a dominant negative effect upon endogenous hTERT in the organelle (Kovalenko and Santos, in preparation).

It can be argued that the mitochondrial dysfunction was a consequence of the senescent phenotype. However, we reject this notion based on the fact that the ALT‐immortalized GM847 cells did not undergo senescence in response to _NES‐hTERT expression; yet they had more dysfunctional mitochondria than controls (Fig. 6). In addition, mtDNA integrity and ROS production were analyzed prior to signs of senescence in the nonimmortalized cells (Fig. 5).

Overall, our findings show that mutations on the NES of hTERT are associated with mitochondrial dysfunction and corroborate a role for hTERT in mitochondrial biology (Santos et al., 2004, 2006a; Ahmed et al., 2008; Haendeler et al., 2009). Further, they are consistent with our previously proposed role for hTERT in mitochondrial quality control (Santos et al., 2004, 2006a).

In vivo telomere elongation function of telomerase is important for protecting nuclear DNA from ROS‐induced damage

We previously demonstrated that on average 75% of nDNA damage was localized at telomeres during replicative senescence (Herbig et al., 2004). In this study, we found that about 50% of DNA damage foci in NHF‐expressing _NES‐hTERT colocalized with telomeric sequences, suggesting that TIF did not result from replication‐based telomere shortening. It is more probable that _NES‐hTERT‐induced TIF and the resulting cellular senescence are a consequence of the inability of the mutant hTERT to extend telomeres. Such telomere dysfunction was likely compounded by high levels of oxidative damage resulting from dysfunctional mitochondria (Fig. 5), ultimately leading the cells to prematurely enter into senescence. Results obtained with the chronic exposure to NAC support this notion (Fig. 5D). Mitochondrially derived ROS damages nDNA as demonstrated by us in both yeast and in mammalian cells (Karthikeyan et al., 2003; Stuart et al., 2006 and this study). It is not surprising that telomeres were a target for oxidative damage given their G‐rich nature and the vulnerability of such DNA bases to ROS (Henle et al., 1999; von Zglinicki, 2002).

It is intriguing that extratelomeric DNA damage was present at such high levels; about 4 lesions per 100 000 bases of DNA are estimated to be present in the nucleus of the mutant (Fig. 4E). This degree of nDNA damage is comparable to the amount of damage incurred in SV40‐transformed fibroblasts submitted to treatment with 100 μm H₂O₂ (Yakes & Van Houten, 1997). The increased ROS generated by faulty mitochondria in _NES‐hTERT is clearly damaging the nDNA to a higher extent than in the controls. However, maintenance of a ‘steady state’ level of lesions is likely compounded by some additional impairment in the mutant. It has been reported that hTERT modulates DNA repair (Masutomi et al., 2005). Based on our results, we can speculate that the DNA repair‐promoting properties of hTERT may require cross talk with other protein(s) that either interact with hTERT through residues in the NES region or that reside in the cytoplasm. Alternatively, hTERT modulation of DNA repair may be associated with its ability to elongate telomeres and the associated promotion of genomic stability, a concept previously proposed by others (Holt et al., 1999; Gorbunova et al., 2002; Sharma et al., 2003; Rubio et al., 2004).

Cell culture

Primary NHF telomerase negative and their respective hTERT derivatives were described and maintained as in Santos et al. (2004, 2006a). Human fibroblasts transformed with the early region of simian virus 40 (GM847) were a kind gift of Dr Robert Weinberg (Whitehead Institute, MIT); cells were maintained as described in Stewart et al., 2002. HeLa cells were obtained from ATCC. Cells were grown in Dulbecco’s modified Eagle high glucose medium (Gibco/Invitrogen, Grand Island, NY, USA) supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin.

Plasmids and transfections

Retroviral pBabe and pLXIN vectors with WT hTERT were described earlier (Santos et al., 2004, 2006a). _NES‐hTERT was generated by site‐directed mutagenesis of WT hTERT in EGFP, pLXIN and pBabe vectors (Santos et al., 2004, 2006a) by replacing leucine residues at position 980 and 987 for alanine (L980A/L987A hTERT). The constructs were sequenced to assure introduction of the mutations. Transient and stable transfections were performed as described previously (Santos et al., 2004, 2006a). Images were acquired using a Zeiss Axiovert 200 fluorescence microscope equipped with ApoTome. The objective lens used was the PlanApo 63×/1.4 oil differential interference microscopy, and the pinhole was set to achieve a z‐resolution of 1.0 μm.

Subcellular fractionations and Western blots

Harvested cells (5 × 10⁷) were washed twice with PBS, centrifuged (4000 g, 5 min) and supernatants were discarded. The pellets were weighed and gently resuspended in HEPES dissociation buffer (5 mm KPO4, pH 7.5, 2 mm MgCl₂, 1 mm 2‐mercaptoethanol), at 9× volume of the wet weight of the cells. Protease inhibitor cocktail (1:100 v/v) was added, and the cells were left on ice for 1 h. Once the cells were swollen (as judged with 5 μL on a slide with a 20/40 objective lens), they were broken by Dounce homogenization for 20–30 strokes until only 0–2 intact cells/microscope field remained. Mannitol–sucrose buffer (2.5×; 0.525 m mannitol, 0.175 m sucrose, 5 mm Tris–HCl, 5 mm EDTA, 5 mm MgCl₂, pH 7.5) was added to a 1× concentration. To collect nuclei, the cells were centrifuged for 5 min at 1600 g (4 °C), the supernatants with remaining mitochondria were collected while the pellet was kept as the nuclear fraction. Supernatants were then centrifuged for 30 min at 12 000 g; the pellets containing mitochondria were washed three times with 1× mannitol–sucrose buffer to remove remaining nuclear DNA and proteins, and then treated with proteinase K (0.05 mg mL⁻¹ final concentration) for 30 min on ice. Protease inhibitors were added, and samples were incubated on ice for an additional 10 min. Mitochondria were then washed three times with 1× mannitol–sucrose buffer. The mitochondrial pellets were resuspended in lysis buffer (1% NP‐40, 0.3 m NaCl, 10% glycerol, 20 mm Tris–HCl, pH 8.0, 14 mm 2‐mercaptoethanol and proteinase inhibitors) and left on ice for 10 min. The lysates were then centrifuged for 2 h at 12 000 g, and the protein was quantified using the Bradford or Lowry assays. hTERT was probed with the anti‐hTERT antibody from Epitomics (1:1000), actin with the antibody available from Millipore and ND1, a mtDNA‐encoded protein, was obtained from Santa Cruz (1:200). The latter was used to normalize the mitochondrial data presented in Fig. 1B. Purity of mitochondrial preparations was probed with antibodies against Ku80 and tubulin; enrichment of the mitochondrial fraction was evaluated based on mtHSP70 (data not shown).

Growth curves

Primary NHF and GM7532 were infected with _NES‐hTERT at PD28 and 15, respectively. Cells were maintained in culture under selection with the appropriate antibiotics (puromycin or G418) for the length of the experiments. Parental cells were cultured as reference controls. Cells were seeded always at same number (50 000) in 75‐ cm² flasks and trypsinized and counted every time they reached 100% confluence, when they were reseeded. Cells were followed for 6 weeks. For chronic antioxidant analysis, we used GM7532 parentals and the derivatives expressing _NES‐hTERT given that expression of the mutant hTERT consistently led to later premature arrest in this cellular background when compared to the NHF strain (see Fig. 2A). The cultured medium supplemented with 5 mm of NAC was replaced every two days. In the case of experiments shown in Fig. 5D, cells were put under NAC after 7 weeks from the time the experiment started. The same results were obtained when _NES‐hTERT was expressed in the pBabe vector (data not shown).

Cell cycle analysis

Primary and GM847 fibroblasts (and respective hTERT derivatives) were serum starved overnight and released from serum starvation by the addition of 10% FBS into the medium for 8 h. Cell cycle analysis by flow cytometry with propidium iodide (PI, Molecular Probes, Carlsbad, CA, USA) was performed as described previously (Santos et al., 2003) using a BD Biosciences FACSCalibur flow cytometer. DNA content analysis was performed by Modi Fit LT (Verity Software House, Topsham, ME, USA) software.

Analysis of senescence by β‐galactosidase activity

NHF and its derivatives were plated on 60‐ mm dishes and stained for senescence‐associated β‐galactosidase activity using the Cell staining kit (Cell Signaling) according to the manufacturer’s protocol.

Telomeric repeat amplification protocol

Hundred nanograms of total protein extracts was assayed for TRAP using TRAPeze kit (Chemicon, Temecula, CA, USA) according to manufacturer’s instructions and with some modifications (Santos et al., 2004).

Southern blots for telomeric length measurements

NHF and its hTERT derivatives were grown under identical culture conditions and harvested for DNA extraction within 8 PDs from the time they were derived. DNA was extracted with phenol:chlorophorm, and mTRF measurements performed using Southern blotting as described in Richards et al., 2008.

Immunofluorescence of DDR, p21 and p16 proteins

Cells were grown on coverslips for at least 48 h prior to immunostaining. After washing with PBS, cells were fixed for 20 min with 4% paraformaldehyde/PBS, permeabilized for 20 min with PBST (PBS containing 0.2% Triton X‐100) and blocked for 1 h with PBS/4% BSA. Cells were incubated with antibodies, diluted in blocking buffer, for 2 h at room temperature, followed by three 5‐ min washes with PBS. The secondary Alexa‐488‐ (Invitrogen) and Cy3‐ (Jackson Immunoresearch, West Grove, PA, USA) conjugated antibodies were diluted 1:1000 in blocking buffer and added to the cells for 1 h in a humidified chamber. After two 5‐ min washes with PBS, cells were mounted using DAPI containing mounting media. Primary antibodies were used at following dilutions: anti‐γH2AX (S139) (Upstate, Billerica, MA, USA), anti‐p21 (C‐19, Santa Cruz, Santa Cruz, CA, USA), anti‐p16 (JC8; Abcam, Cambridge, MA, USA), 1:1000; anti‐ATM (S1981) (Abcam), 1:200; anti‐Chk2 (T68) (Lot1; Cell Signaling), 1:500; anti‐p53(S15) (Cell Signaling, Danvers, MA, USA), 1:100. Images were acquired using a Zeiss Axiovert 200 fluorescence microscope equipped with ApoTome.

ImmunoFISH

Cells were processed and stained with anti‐γH2AX antibodies as described earlier. Following three 5‐ min washes with PBS, the cells were incubated with Alexa‐488 conjugated rabbit anti‐mouse antibodies diluted 1:1000 in PBS for 1 h. After two 5‐ min washes with PBS, the cells were incubated with 4% paraformaldehyde/PBS for 20 min, washed 2 × 5 min with PBS and dehydrated by sequentially placing them in 70%, 90% and 100% ethanol for 3 min each. Nuclear DNA was denatured for 5 min at 80 °C in hybridization buffer containing 0.5 μg mL⁻¹ (C₃TA₂)₃‐Cy3‐labeled peptide nucleic acid (PNA), 70% formamide, 12 mm Tris‐HCl (pH8), 5 mm KCl, 1 mm MgCl₂, 0.001% Triton X‐100 and 2.5 mg mL⁻¹ acetylated BSA. After denaturation, incubation was continued for 2 h at room temperature in a humidified chamber. Cells were washed two times for 15 min with 70% formamide/2 sodium chloride/sodium citrate (0.3 m NaCl, 30 mm Na‐ Citrate (pH 7), followed by a 10 min wash with 2× SSC and a 10 min wash with phosphate buffered saline plus Tween. To reinforce the protein signal, cells were incubated for 1 h with an Alexa‐488 conjugated goat anti‐rabbit secondary antibody. Cells were mounted as described earlier and analyzed by immunofluorescence microscopy using a Zeiss Axiovert 200 fluorescence microscope equipped with ApoTome. Images were acquired as z‐stacks spaced 0.4 μm apart using a 100× lens with 1.4 optical aperture.

DNA integrity analysis

Nuclear and mtDNA damage were estimated using QPCR as described previously (Santos et al., 2002, 2006b; Kovalenko & Santos, 2009). Basal levels of damage were analyzed using at least two independently derived cell lines. Two independent DNA extractions were performed for each cell type. The DNA was probed twice for each set of genes analyzed to assure reproducibility of the results using the same DNA. Levels of lesions are calculated relative to the amount of damage detected in the non‐hTERT control as previously described (Santos et al., 2002).

Mitosox measurements

Superoxide levels were monitored by confocal microscopy and by FACS based on MitoSox Red (Invitrogen) fluorescence. For FACS analysis, cells were loaded with 1 μm of Mitosox diluted in Hank’s buffered salt solution in tissue culture flasks and incubated at 37 °C for 10 min. Fluorescence levels were calculated based on the average results obtained for 10 000 cells per cell type. Experiments by FACS were independently repeated three times. For confocal analysis, cells were plated on glass coverslips 48 h prior to the experiment, incubated at 37 °C for 10 min with 1 μm of Mitosox diluted in HBSS buffer. After incubation, the fluorescent indicator was removed and cells were washed with warm HBSS buffer. A control using 5 μm of the mitochondrial uncoupler FCCP was included to assure specificity of the signals. Images were acquired using laser scanning confocal microscope (Radiance 2000; Bio‐Rad, Hercules, CA, USA). At least ten images were taken, and approximately 50 cells were analyzed/coverslip for each cell type. Approximately 300 cells per cell type were analyzed in three to five independent experiments.

Oxygen consumption

Oxygen consumption was monitored as described previously (Kristian et al., 2006). We performed experiments in intact cells using a closed chamber and a Clark‐type oxygen sensor (Hansatech Instruments, Norfolk, England). The same number of cells was analyzed for each cell type in triplicate experiments. The rate of respiration was calculated based on the ratio between the basal rate and the maximal rate of oxygen consumption in the presence of FCCP, an uncoupler of the mitochondria. The rate of proton leak was estimated by dividing basal O₂ consumption in the presence of oligomycin (5 μg mL⁻¹), the mitochondrial ATP synthase inhibitor alone (pseudo state 4) by the maximal rate of O₂ consumption in the presence of FCCP (5 μm) and oligomycin.

Electron microscopy

NHF and its hTERT derivative were processed in sections (approximately 1 mm³), which were rapidly fixed in diluted Karnovsky’s fixative (5% Glutaraldehyde, 4% formaldehyde in 0.08 m sodium phosphate buffer). Embedded sections (0.5 μm) were cut with a glass knife and stained with Toluidine blue for orientation. Ultrathin sections were cut with a diamond knife, stained with uranyl acetate and lead citrate and viewed on a Philips Morgagni electron microscope (Philips, Amsterdam, Netherlands). Structurally damaged mitochondria were operationally defined as having loss or dissolution of ≥ 25% of cristae, alteration of size and vacuolization as judged by a pathologist. The same number of cells, from independent slides, were evaluated.

Statistical analysis

Student’s t‐test and one‐way anova were used to calculate statistical significance as described in figure legends.