Mitochondrial Telomerase Protects Cancer Cells from Nuclear DNA Damage and Apoptosis

1Telomerase is an enzyme best known for its role in telomere maintenance. Cells with low or no telomerase expression lose telomere repeats during cell division, eventually resulting in cellular senescence. Most cancer cells, germ cells and embryonic stem cells express high levels of telomerase, thus contributing to pluripotency and immortality. In order to maintain telomeres the enzyme needs its catalytic subunit (TERT) as well as the RNA component (TERC or TR) which contains the template for telomere synthesis.

2In recent years, however, evidence has accumulated that telomerase, and in particular its catalytic subunit TERT, is involved in various non-telomere-related functions such as regulation of gene expression, growth factors and cell proliferation –. In addition, various groups have shown that TERT shuttles from the nucleus and translocates to mitochondria upon exogenous stress –. We and others have shown a protective role of telomerase within mitochondria – while inability of telomerase shuttling leads to cellular stress, prevents immortalization and increases sensitivity against genotoxic stress, as shown recently by Santos’ group –.

3Most cancer cells express high levels of telomerase, an important prerequisite for indefinite proliferation and immortality. In addition, telomerase contributes to tumorigenesis via non-telomere dependent mechanisms which are not well understood yet.

4Telomerase has thus been suggested to be an important anti-cancer target, with the first clinical trials of the telomerase inhibitor imetelstat successfully under way,. Telomerase is regulated at multiple levels and subcellular localization is one of them. Cancer cell survival after therapeutic treatments can be heterogeneous with some cells responding to the treatment while others seem to be resistant, contributing to tumor cell survival. A better insight into the biological consequences of different subcellular localizations of TERT might lead to the development of more effective anti-cancer treatments.

5Here we characterized the exclusion of telomerase from the nucleus upon stress application and found a heterogeneous stress response in cancer cell populations. Importantly, there was a striking correlation between telomerase/TERT retained within the nucleus and high DNA damage. In contrast, cells which excluded telomerase quickly from the nucleus accumulated no or very low amounts of DNA damage. By modeling the different subcellular localizations of telomerase using organelle-targeted “shooter” vectors we demonstrate here that mitochondrial telomerase prevents nuclear DNA damage as well as the induction of apoptosis after treatment with H2O2 and irradiation. We suggest that reduced generation of mitochondrial reactive oxygen species (ROS) could be the underlying mechanism to explain how mitochondrial TERT prevents nuclear DNA damage.

6Thus, exclusion of telomerase from the nucleus after stress, such as anti-cancer therapeutic treatment could be a protective mechanism that decreases nuclear DNA damage and apoptosis by reducing oxidative stress within mitochondria. This might contribute to increased resistance of those cancer cells against various anti-cancer treatments.

7Subcellular shuttling of TERT protein from the nucleus to mitochondria had been shown previously in various cell types, including cancer cells, –. We confirmed this shuttling of endogenous telomerase after H2O2 treatment in HeLa and MCF7 cells (Fig. 1A) from the nucleus to mitochondria and quantified the exclusion compared to MRC-5/hTERT cells (Table 1). In order to evaluate the shuttling kinetics of TERT in more detail we analyzed 3 cell lines, including 2 cancer cell lines as well as hTERT over-expressing MRC-5 fibroblasts, and followed them over 5 days.

8In all 3 cell lines nuclear TERT exclusion started around 45 min after onset of H2O2 treatment. In hTERT over-expressing fibroblasts the exclusion reached its maximum of 60% after 3 hours while it took both cancer cell lines up to a day to reach an exclusion level of 50–60% (Fig. 1B–D). This corresponds well with the TERT mitochondrial co-localization data from our confocal images from the 3 cell lines (Table 1). Intriguingly, the nuclear exclusion level of 50–60% persisted in all 3 cell lines up to 5 days, the longest time points we analyzed (Fig. 1B–D and S1). Thus, nuclear TERT exclusion is a rather persistent process that can last up to several days after a single bolus dose of 400 µM H2O2. To our knowledge, this is the first time that the TERT exclusion kinetics has been investigated in detail and compared in three different cell lines over a 5 day time frame. The long persistence of TERT protein outside the nucleus in the cancer cell lines might be an important contributor to increased resistance and decreased apoptosis in cancer cells after drug treatment or irradiation compared to non-cancer cells which in most cases express no or rather low levels of telomerase. We have shown previously that telomerase negative fibroblasts are much more susceptible to apoptosis after treatment with H2O2 and etoposide than their telomerase over-expressing counterparts. We had also shown previously that TERT exclusion from the nucleus is reversible over a time frame of around 10 days. However, we do not know whether the persistent TERT protein within mitochondria does always come from the nucleus or whether newly synthesized TERT protein is directly imported into mitochondria. This question requires further investigation.

9Next we correlated telomerase exclusion for each individual cell with its DNA damage level at 3 hours after H2O2 treatment. We quantified the TERT exclusion level for each single cell as either nuclear, if more than 75% of total TERT was in the nucleus, or cytoplasmic/mitochondrial if more than 75% of TERT was outside the nucleus with the remaining cells being classified as an intermediate phenotype.

10We found a clear heterogeneity for nuclear TERT exclusion between cells within a population (Fig. 2A and S2A). Intriguingly, we found that cells that had excluded telomerase 3 hours after the treatment showed no or very low DNA damage whilst those with nuclear telomerase had a significantly higher amount of nuclear DNA damage in all 3 cell lines (Fig. 2). Also, cells with an intermediate exclusion pattern showed an intermediate damage level, still significantly higher than cells with completely excluded telomerase but not significantly different from those with predominantly nuclear telomerase. This data suggests that a high level of nuclear exclusion (>75%) and mitochondrial localization of TERT is required in order to exert its protective function –. Absolute DNA damage levels were different between the 3 cell lines with the two cancer cell lines displaying much higher damage levels than TERT over-expressing fibroblasts. We also measured the absolute TERT signal intensities for the 3 different localizations and found that mitochondrial TERT signal was always lower than that in the nucleus or in the intermediate state (Fig. S2B). It had been shown previously that TERT protein level is down regulated within mitochondria after H2O2 treatment which could explain this observation.

11A correlation between higher DNA damage in cancer cells and nuclearly confined telomerase unable to shuttle due to a mutation in its nuclear export signal was described recently by Kovalenko and colleagues. The mutated TERT induced an increase in spontaneous telomeric as well as non-telomeric nuclear DNA damage in 2 cancer cell lines compared to the same cells without the mutant TERT. In addition, cancer cells with a mutant TERT that was confined to the nucleus and unable to shuttle lost their proliferation ability, were not able to form colonies in soft agar and showed an increased amount of mitochondrial DNA damage. Together, these results suggest that sub-cellular shuttling of TERT might have important implications for the sensitivity of cells against DNA damage. This increased resistance due to high telomerase expression and nuclear exclusion of TERT might favor the survival of cancer stem cells which could result in relapse after therapy. There is also previously published data of a protective role of nuclear TERT against staurosporine induced apoptosis. However, staurosporine is a protein kinase inhibitor that activates apoptosis in a rapid manner without inducing DNA damage and independent from mitochondria. We therefore suggest a different mechanism of action in both experiments.

12We next modeled the different TERT localizations separately by over-expressing nuclear and mitochondrial organelle specific vectors expressing the catalytic telomerase subunit TERT fused to a myc-tag in 3 cancer cell lines: HeLa, MCF7 and U87 glioblastoma (Fig. 3). Transiently transfected cells were treated either with H2O2 or irradiation and analyzed for γH2A.X DNA damage foci and TERT localization using the fused myc-tag. The localization for mitochondrial (mito TERT) and nuclear TERT is shown in Fig. 3A and B (upper panels). There was no difference in DNA damage levels between cells transfected with either vector or un-transfected cells before treatment. However, we found that in all 3 cell lines and both treatments, cells with a mitochondrial TERT localization had significantly less DNA damage compared to those cells that either expressed nuclear TERT or were un-transfected (Fig. 3 A–D). In order to exclude that endogenous telomerase interacted with the over-expressed shooter TERT protein we repeated the experiment in an SV40 immortalized MRC-5 cell line that maintains its telomeres via an alternative lengthening mechanism. Post irradiation, we found the same protective effect of mitochondrial telomerase (Fig. 3 E). We also used an antibody against 53BP1, another protein involved in the DNA damage response to confirm our results that DNA damage foci are high in cells transfected with nuclear TERT while in contrast those transfected with mitochondrial TERT show less DNA damage than un-transfected cells (Fig. S3).

13Since high amounts of nuclear DNA damage are thought to decrease survival of cells we analyzed whether the used stress treatments would also compromise cell survival and induce apoptosis. We treated 3 cell lines transfected with both TERT shooter vectors with H2O2 and X-irradiation and determined apoptosis induction using an antibody against activated caspase 3. Intriguingly, not a single cell transfected with mitochondrial TERT showed any sign of apoptosis while around 20% of un-transfected cells and between 40–60% of cells expressing the nuclear shooter were apoptotic (Fig. 4). This result confirms that indeed the induced DNA damage found in cells with nuclear TERT localization impacts directly on cell survival while mitochondrial TERT efficiently protects against apoptosis. In order to elucidate the mechanism by which mitochondrial TERT might protect cancer cells from nuclear DNA damage we measured the amount of mitochondrial ROS after H2O2 treatment and irradiation using mitosox staining as a measure of mitochondrial superoxide generation in addition to myc-TERT staining for nuclear and mitochondrial “shooter” vectors in the same 3 cancer cell lines (Fig. 5 A–E). Mitosox dye is taken up by mitochondria in a membrane potential specific manner. In order to exclude that different membrane potential caused the effects, we measured mitochondrial membrane potential in HeLa and MCF7 cells transfected with both TERT shooters and compared them to un-transfected cells after H2O2 treatment as well as X-irradiation. In accordance with our previous findings of an increased mitochondrial membrane potential in MRC-5/hTERT compared to parental fibroblasts the results confirm a significantly higher membrane potential in cells over-expressing mitochondrial TERT, which is in HeLa cells already apparent even before any stress treatment (Fig. S4). Therefore, mitosox levels found in our experiments truly represent different ROS levels which are dependent on TERT localization. We found that over-expression of mitochondrial TERT in all cell types resulted in significantly lower ROS levels after H2O2 treatment and irradiation compared to un-transfected cells or those that over-expressed nuclear TERT (Fig. S4).

14Again we used MRC-5/SV40 cells without endogenous telomerase to confirm the results from the 3 cancer cell lines and found the same protective effect of mitochondrially localized TERT on ROS levels (Fig. 5F). ROS levels in cells expressing the nuclear TERT “shooter” were usually not different from un-transfected cells with the exception of MCF7 and MRC-5/SV40 cells after irradiation, where nuclear shooter transfected cells showed lower ROS levels than un-transfected cells.

15This data suggests that by shuttling into mitochondria telomerase/TERT not only protects the organelle, but by decreasing mitochondrial superoxide production also indirectly protects the nucleus from DNA damage. Similar observations of telomerase shuttling from the nucleoplasm to nucleoli had been reported previously under ionizing radiation in primary and cancer cells. The authors had speculated that telomerase might negatively interfere with repair enzymes in the nucleus under conditions of increased DNA damage and stress. Our results seem to support this suggestion that telomerase might be “undesirable” within the nucleus under conditions of DNA damage. Telomerase has been shown to “heal” chromosomes by attempting a form of DNA repair by adding telomere sequences to broken telomere ends. However, this might not lead to proper DNA repair, as known to be carried out by true DNA repair systems.

16Our results demonstrate that mitochondrial telomerase localization specifically decreases mitochondrial ROS generation and cellular oxidative stress after induction of exogenous stress generated by H2O2 or irradiation in cancer cells and might thereby prevent damage to nuclear DNA. It could also explain why shuttling of telomerase from the nucleus to mitochondria seems to promote cellular survival, whilst in cells where telomerase is not able to leave the nucleus DNA damage accumulation is observed.

17Diehn and colleagues reported recently that cancer stem cells that produced less ROS due to higher antioxidant expression accumulated less DNA damage after ionizing radiation. It would be interesting to ascertain whether these cells also have more excluded telomerase than non-cancer stem cells after irradiation. It has been shown that cancer stem cells express high telomerase activity level; however nothing is known about TERT shuttling in these cells.

18We have demonstrated previously that exogenous ROS generation by irradiation in fibroblasts damages mitochondria and accelerates nuclear DNA damage creating a positive feedback loop. We suggest that such a functional interaction between mitochondria and the nucleus also exists in cancer cells and other telomerase positive cells, where telomerase enters mitochondria in order to decrease ROS which are induced by chemotherapeutic drugs and irradiation. Thus, it seems that anti-cancer treatments can induce a novel, hitherto unknown mechanism of telomerase shuttling that prevents nuclear DNA damage by decreasing mitochondrial ROS generation via induction of telomerase shuttling. Due to its heterogeneous pattern it could also explain the resistance of some, but not all, cancer cells against therapeutic treatments.

19HeLa, MCF7, U87 and MRC5/SV40 cells originated from ATCC. MRC5 were purchased from ECACC (London). hTERT overexpression was performed using retroviral transfection of hTERT as described earlier. U 87 and MRC5/SV40 cells were cultivated in MEM and DMEM respectively supplemented with 1% non-essential amino acid, 10% FCS (Sigma), 2 mM glutamine, and 1% penicillin/streptomycin. All other cell types were cultivated in DMEM (PAA) containing 10% FCS (SIGMA), 1% penicillin/streptomycin (PAA) and 2 mM glutamate (Gibco). Cells were incubated at 37°C at ambient oxygen and 5% CO2.

20Cells were seeded 1 or 2 days before treatment onto 19 mm coverslips in 12 well plates for immuno-fluorescence staining (5×104 per well).

21H2O2 (SIGMA) treatment was performed in serum free medium for the indicated times and concentrations. The used H2O2 concentrations had been optimized for each different type of experiment. X-irradiation at 10 and 20 Gy was performed using a Faxitron (Elektron Technology, UK). For all irradiation experiments (except apoptosis analysis) cells were fixed and analyzed 20 minutes after treatment. Cells were fixed using 4% paraformaldehyde in PBS for 10 min.

22Cells were fixed on coverslips using 4% paraformaldehyde for 10 min. Single or double immuno-staining was performed with the following primary antibodies: mouse γH2A.X (Upstate), rabbit anti-TERT (Rockland), rabbit anti-Ki67 (Abcam) and anti myc-tag (Abcam). Specificity and lack of background staining of the used TERT antibody has been confirmed (Fig. S5). Secondary antibodies were: goat anti-mouse and rabbit Alexafluor 594 and 488 (Molecular probes/Invitrogen). Nuclear staining was observed using DAPI. Images for each channel were obtained using an AxioImager Z1 microscope (Zeiss) equipped with suitable filter cubes to spectrally distinguish Alexafluor488 and Alexafluor594, ensuring no bleedthrough into/from either fluorophore (established previously, data not shown). Images were subsequently analysed using ImageJ. Thresholds were defined for each image individually.

23Cells were loaded with 400 nM mitotracker green (Molecular probes) for 30 minutes at 37°C before fixation and then stained with anti-TERT antibody and Alexa Fluor® 594. Images were captured using a Zeiss LSM 510 equipped with a 63×1.4 NA objective with the pinhole set to 1 Airy unit (Zeiss, Germany). Z stacks were obtained every 100 nm for each cell (sampling >2×Nyquist criteria). Images were deconvolved and then analyzed for object co-localization in 3D using Huygens software (SVI, Netherlands). For colocalization analysis and image preparation, the deconvolved images were rendered into 3 dimensional volumes. Single cells were isolated within the image and single, unified objects (mitochondrial and Tert) were identified using 'Object Colocalization' in Huygens with a cut-off to exclude any objects smaller than one mitochondrion in the mitotracker green channel. Huygens then computed a Pearson's correlation coefficent for each cell for 3 dimensional colocalization between mitochondria and Tert staining. This enabled us to accurately predict true colocalization (within the limits of diffraction limitation) of the two dyes irrespective of their different wavelengths.

24TERT containing nuclear and mitochondrial organelle specific vectors (pCMV-myc-mitoTERT and pCMV-myc-nucTERT were a kind gift from J. Haendeler and Joachim Altschmied, Duesseldorf, Germany and described previously, ). Transient transfection was performed using lipofectamine ™ 2000 (Invitrogen, USA) of mito-TERT and nuc-hTERT “shooter” vectors. The average transfection efficiencies 48 hours after transfection were between 25 and 30%. Transiently transfected cells were treated 2 days after transfection either with H2O2 or irradiation.

25Cells were stained with 5 µM mitosox (Invitrogen, USA) for 15 min after H2O2 treatment or irradiation prior to fixation and antibody staining. ROS levels were determined as the percentage of cytoplasmic mitosox signal from total cytoplasmic area using Image J (http://rsbweb.nih.gov/ij/).

26Analysis of DNA damage was performed using immuno-fluorescence either as a single staining with γ-H2A.X or double staining with TERT. Cells were fixed, permeabilized with PBG (PBS, BSA, fishskin gelatine and 0.5% triton) and γ-H2A.X antibody was applied to the cells and then stained with Alexa Fluor® 594. Anti myc-tag and Alexa Fluor® 488 were used to visualise transfected TERT protein. Slides were examined using a Zeiss Axiovision fluorescence microscope (Zeiss, Germany). For the analysis of DNA damage the number of DNA damage foci was counted for each type of TERT localization separately from 20–40 cells per group and cell line.

27For each individual cell, TERT localization was manually quantified by comparing telomerase signals inside and outside the nucleus using Image J. Subcellular areas were determined for nuclear and cytosolic regions by using freehand selection. Expression signals in the selected area were evaluated using area calculation function after thresholding to remove noise. The result of each individual cell indicated a percentage of TERT signal expressed in the subcellular compartment:

28% TERT nuclear expression = TERT signal in nucleus area/total TERT signal×100,

29% TERT cytoplasmic expression = TERT signal in cytosolic area/total TERT signal×100. The average percentage of nuclear and cytoplasmic localization of TERT from at least 30 individual cells was taken to determine the average percentage of the whole population.

30We classified the localization of TERT into 3 classes: nuclear TERT (N): 75%–100% of TERT signal resides within the nucleus, cytoplasmic TERT (C): 75%–100% of TERT signal resides outside the nucleus and all other percentages for the class of intermediate localization (I). For each of the 3 classes we determined the number of γH2A.X foci from around 40–100 cells per cell line.

31Hela, U87 and MRC/SV40 cells were transfected with nuclear and mitochondrial TERT shooter. After 2 days they were treated either with 400 µM H2O2 or 20 Gy irradiation and left for 1 more day (U87 and MRC/SV40) or 2 days for HeLa due to a known delay in apoptosis induction in these cells. After fixation cells were stained with myc-tag for TERT and activated caspase 3 (Abcam) in order to label apoptotic cells. Results have been determined from 30–150 transfected cells per cell line and condition.

32One way ANOVA was performed using Sigma Plot (Systat Software Inc, USA).