ATM-dependent phosphorylation of SNEV^hPrp19/hPso4 is involved in extending cellular life span and suppression of apoptosis

Oxidative stress induces an additional SNEV protein band in Western blotting

From previous results we know that SNEVexpression is induced upon oxidative and DNA DSB damage [23,26]. To further study the effect of oxidative stress on SNEV, we treated HeLa cells with 1mM hydrogen peroxide for different times and detected additional bands above the expected 56 kDa band in a Western Blot with anti-SNEV antibody, and band intensity increased in a time-dependent manner. These higher bands were removed after dephosphorylation using calf intestinal phosphatase (Fig. 1A). Similarly, this additional SNEVband appeared when we treated human diploid fibroblasts (HDF5) with bleomycin (Fig. 1B).

These results prompted us to ask if these bands might represent a phosphorylated form of SNEV in response to oxidative stress. Therefore, we isolated SNEV as GFP-fusion from stable HeLa cells constitutively expressing GFP-tagged SNEVunder the control of the endogenous promoter as described in [32] and submitted it to mass spectrometric analysis. The advantage of using GFP-SNEV-Hela lies in allowing the use of a commercially available GFP-antibody for IP which is covalently coupled to agarose beads, thus avoiding predominant detection of immunglobulins in the mass spectrometric analysis. In order to make sure that GFP-SNEV behaves similarly to SNEV in cells we confirmed its localization to nuclei (Fig. S1A) as well as its ability to integrate into the SNEV/CDC5L complex by co-precipitation of SNEV and CDC5L with GFP-SNEV by GFP antibodies (Fig. S1B).

Interestingly, S149 also represents an ATM target site (Fig. 1D). Fig. S2 shows that the signal detected by MS is specific for H₂O₂-treated cells. This is not due to a difference in the efficiency of the IP, since the Western Blot in fig. S1B clearly shows that comparable amounts of GFP-SNEV were precipitated under both conditions. Table S1 gives a list of the identified phosphopeptides of SNEV in the H₂O₂-treated sample with site probability calculated by phosphoRS 2.0.

Taken together, the mass spec spectra, the presence of an ATM kinase consensus site, and ATM's known function as a kinase signaling oxidative stress prompted us to generate an anti-pSNEV(S149) specific antibody.

Phosphorylation of SNEV at S149 is induced by oxidative stress in different cell types

In order to take a closer look at this possible phosphorylation site of SNEV, a phospho-specific antibody was generated. This anti-pSNEV(S149) antibody detects a band at the expected size of SNEVupon oxidative stress treatment of Hela cells, and this signal is abolished upon phosphatase treatment (Fig. 2A). Besides the bands at the expected size, other bands are as well detected by anti-pSNEV(S149) antibody on Western Blot of Hela lysates (Fig. S3), but at higher molecular weights. These bands might represent post-translationally modified SNEV in complexes similar to the ones described in [24].

Based on these results, we were curious to confirm our assumption that SNEV phosphorylation was stimulated by DNA damage in normal cells. Human diploid fibroblasts (HDF5) were exposed to different concentrations of hydrogen peroxide for 1 hour or to 100μM hydrogen peroxide for different times. Total cell lysates were subjected to Western Blot analysis with anti-pSNEV(S149). Indeed, phosphorylation of SNEV was triggered by oxidative stress in a time- and dose-dependent manner (Fig. 2B and C). Phosphorylation was observed as quickly as after 5 minutes of hydrogen peroxide treatment and reached a maximum after 1 hour. To verify that hydrogen peroxide treatment indeed induced the formation of DNA double strand breaks (DSB), we examined the ATM dependent phosphorylation of γH2AX, which occurs in response to DSB [7], and showed a similar kinetic as phosphorylation of SNEV(Fig. 2B and C).

Different types of DNA damage induce nuclear localization of pSNEV

In order to localize pSNEV within the cells, we first tested for the specificity of the antibody for indirect immunofluorescence studies. Using either the phospho-peptide that was used for generation of the antibody or a non-phospho-peptide with the same sequence, for pre-blocking the anti-pSNEV antibody, staining was abolished by blocking with the phospho-peptide, while no difference was seen after pre-incubation with the non-phospho-peptide (Fig. S4).

In addition to a strong and predominant nuclear signal upon oxidative stress, the phosphorylated form is also detected in the cytoplasm (Fig, 3A), and it is currently unclear if this phosphorylation is a signal for shuttling of SNEV between nucleus and cytoplasm. In any case, under normal conditions only very few cells stained positive for phosphoSNEV, which seems to localize to the cytoplasm (Fig. 3A).

To test if phosphorylation of SNEVis restricted to oxidative stress conditions or if it is a general step in the DNA damage response, we challenged cells with mitomycin C (MMC), a potent DNA crosslinker. HeLa cells were incubated with 25μg/ml MMC for 24 hours and submitted to indirect immunofluorescence staining with anti-pSNEV(S149). As observed for hydrogen proxide, also MMC induces phosphorylation of SNEV, and the phosphorylated species is mainly detected in the nucleus (Fig. 3B).

Thus, we assume that the phosphorylation of SNEVis an integral part of the nuclear DNA damage response.

Phosphorylation of SNEV at serine 149 is ATM-dependent

In order to test if this phosphorylation is dependent on ATM, we took advantage of the reagent 2-(4-Morpholinyl)-6-(1-thianthrenyl)-4H-pyran-4-one (KU-55933), a specific inhibitor of ATM kinase activity [33].

Indeed, we observed markedly reduced levels of phosphorylated SNEV upon oxidative stress, when ATM was blocked by Ku-55933. Similarly, upon oxidative stress we detected SNEV as a double band with anti-SNEV antibody, and Ku-55933 treatment abolished the upper band, supporting the idea that this upper band represents phosphoSNEV (Fig. 4A). The fact that the phosphoSNEV bands did not disappear completely could be due to another kinase that backs up ATM function. It is known that both ATM and ATR can phosphorylate H2AX [34], which also explains that H2AX phosphorylation is only slightly reduced upon ATM inhibition.

To confirm that ATM controls SNEV S149 phosphorylation with an independent method, we used mouse embryonic fibroblasts in which ATM was flanked by loxP sites and cut out using Cre recombinase (ATM -/- MEF). As shown in fig. 4B, the induction of pSNEV(S149) upon hydrogen peroxide treatment was reduced to less than half of that observed in wildtype MEF (pSNEV band intensity 0.16 versus 0.38 arbitrary units, normalized to βActin intensity; data not shown). The lower panel in fig. 4B shows an agarose gel of the genotyping PCR. Using a combination of three primers, three different fragments are amplified according which cells can be genotyped. The 320 bp fragment indicating absence of loxP sites is amplified only in wildtype cells, whereas presence of loxP sites gives rise to a 420 bp fragment. Only after introduction of Cre recombinase is ATM cut out and a fragment of 920 bp is amplified. Since the 420 bp fragment is still present inspite of Cre transfection, there is still residual ATM activity in our ATM -/- MEF, which explains why we still detect some pSNEV in these cells, although to a lesser extent then in wt MEF.

Taken together, this suggests that SNEV is phosphorylated upon DNA damage in an ATM-dependent manner and the presence of the ATM target motif suggest that it might be a novel direct substrate for ATM.

Effect of phosphorylation on life span and stress resistance in HUVEC

To investigate the physiological effect of the phosphorylation at S149 of SNEV, we generated stable HUVEC cell lines overexpressing either wildtype SNEV (wtSNEV HUVEC) or a point mutant SNEV where the ATM target site was abolished by exchanging serine 149 for alanine (SNEV S149A HUVEC). Both recombinant cell lines showed comparable overexpression of SNEV versus untransfected HUVEC (Fig. 5A). Overexpression was additionally confirmed by IF (Fig. S5).

Using these cell lines, we studied how the lack of phosphorylation at S149 affects DNA damage levels, stress resistance and life span.

As described earlier [26], overexpression of wtSNEV had a strong life-span extending effect. Indeed, SNEV wt HUVEC underwent 30 population doublings post transduction, while empty vector control HUVEC reached senescence already upon PD 8 post transduction, thus SNEVwt overexpression more than tripled the replicative life span. Surprisingly, SNEV S149A overexpression did not fully abrogate this effect of SNEV and also extended the replicative life span of HUVEC, albeit to a significantly lower extent of around 18 PD post transduction (Fig. 5B). Replicative senescence at the respective replication end points was confirmed by staining for the widely accepted senescence marker SA β-galactosidase expression (data not shown).

Since we bserved phosphorylation of SNEV upon treatment with DNA damaging agents, we hypothesized that phosphorylation might be necessary to elicit a proper DNA damage response. Therefore we performed comet assays to test for DNA integrity under basal conditions.

Indeed wtSNEV overexpressing cells were mainly assigned to categories I or II, corresponding to no or low levels of DNA damage. In contrast, the SNEV S149 HUVEC have lower levels of DNA damage than empty vector controls, but still increased damage levels compared to wtSNEV HUVEC. We conclude that increased SNEV levels lead to increased DNA maintenance in unstressed conditions, while the full effect of more efficient repair unfolds only if SNEV can be phosphorylated at S149.

In addition to the basal levels of DNA damage, we assessed the cellular response to treatment with genotoxic and/or oxidative stress inducing reagents. Therefore, we used combined BSO/bleomycin treatment as described earlier [26]. WtSNEV HUVEC did not enter apoptosis upon the treatment in accordance to our earlier results [26], which is also in agreement with findings by the Legerski lab, that GFP-SNEV overexpression in Hela cells reduces MMS induced apoptosis by two-fold [24]. In contrast, in SNEV S149A HUVEC the BSO/bleo-mycin treatment clearly induced apoptosis/necrosis. The initially low level –when compared to empty vector control cells- of cell death of 10% for SNEV S149A HUVEC more that doubled upon treatment to 22.5%.

Since SNEV is required for repair of DNA interstrand cross-links [25], we tested if SNEV overexpression protected cells from the deleterious effects of the DNA cross-linking agent cisplatin. Figure 5E shows that cisplatin treatment very strongly induced apoptosis in empty vector control HUVEC as well as in SNEV S149A HUVEC, while wtSNEV HUVEC were clearly more resistant to this treatment. Therefore, we conclude that phosphorylation of SNEV at S149 is necessary for efficient corsslink-repair activity of SNEV.

In summary, we conclude that overexpression of S149-phosphorylation incompetent SNEV does have a small but discernible effect on the DNA damage status under normal culture conditions when compared to wtSNEV overexpression, but does not exert the strong anti-apoptotic effect of wtSNEV after induction of DNA damage by either oxidative stress or interstrans crosslinks.

In any case, the life-span extending and DNA repair functions of SNEV do partially depend on phosphor-rylation at S149, most probably in dependence of ATM.

Cells and culture conditions

HeLa cells were grown in RPMI1640 (Biochrom AG, Berlin, Germany) supplemented with 10% fetal calf serum (FCS).

Human diploid fibroblasts (HDF) were grown in DMEM/HAM's F12 1:2 (Biochrom AG, Berlin, Germany) supplemented with 10%FCS.

Human umbilical vein endothelial cells (HUVEC) were cultivated in EBM Basal Medium plus EGM SingleQuot Supplements & Growth Factors (Lonza, Basel, Switzerland) supplemented with FCS to a final 10%. For HUVEC, flasks were precoated with 0.1% gelatine in PBS.

GFP-SNEV-Hela were established by transfecting Hela cells with a SNEV-BAC construct containing all regulartory elements obtained from Ina Poser, Max Planck Institute for Molecular Cell Biology and Genetics, Dresden, Germany. GFP-SNEV-Hela were cultivated in RPMI1640 (Biochrom AG, Berlin, Germany) supplemented with 10% fetal calf serum (FCS) and 800 μg/ml G418 (Gibco, now life technologies, Grand Island, NY, USA).

Confluent adherent cultures were detached using 0.1% trypsin and 0.02% EDTA and were passaged with an appropriate split ratio of 1:2 or 1:4 once or twice a week depending on confluence and population doubling level (PD). Subsequently, cumulative PDs were calculated as a function of passage number and split ratio [47].

Fibroblasts (HDF5) 2x10⁵ /well were seeded in 6 well plates. The next day, cells were treated with H₂O₂ (100, 200, 400 μM/ml) for 1 hour and 100μM/ml for different time spans (5, 15, 30, 60, 120 minutes).

Fibroblasts (HDF5) were seeded in 6 well plate at a density of 2x10⁵ cells/well 24 hours before treatment with 2.5, 5, 10, 25, 50, 100μg/ml of Bleomycin for 1 hour and 25μg/ml for (5,15,30,60,120 minutes). For Mitomycin C treatment, fibroblasts cells were exposed to different concentration (2.5, 5, 10, 25, 50 μg/ml) for 24 hours. Cell were scraped directly on ice in 2x SDS loading dye, sonicated and boiled. Proteins were separated by SDS gel electrophoresis in 4-12% BT gel (Invitrogen, now life technologies, Grand Island, NY, USA).

To induce apoptosis, cells were seeded into 12-well cell culture plates at a density of 10000 cells /cm². To enhance the sensitivity of cells to bleomycin-induced apoptosis, cells were cultivated in the presence of 1 mM buthionine-sulfoximine (BSO, Sigma-Aldrich, St.Louis, MO, USA) for 48 h before treatment with with 100 μg/ml of bleomycin (Sigma-Aldrich, St.Louis, MO, USA) for further 24 h. Alternatively, cells were treated with 100 μg/ml of cisplatin (Sigma) for 24h to asssay apoptosis induction by DNA cross-links.

ATM deletion in mouse embryonic fibroblasts

Primary mouse embryonic fibroblasts (MEF) were prepared from E13.5 fetuses (ATM^flox/flox). Cells were cultured in DMEM high glucose medium (Sigma-Aldrich) supplemented with 15% fetal bovine serum (Lonza), 1% L-glutamine (Gibco, now life technologies, Grand Island, NY, USA), 1% penicillin and strepromycin, 1% NEAA (non essential amino acids, Gibco), 1% sodium pyruvate (Gibco) and β-mercaptoethanol (1:140000).

Large T-Antigen was used to immortalize MEFs (ATM^flox/flox). For the generation of retrovirus, Large T-Antigen -MSCVneo plasmid was co-transfected with the packaging plasmid pCL-Eco and the enveloping plasmid VSV-G in 293T cells. Supernatant was collected after 24 and 48 hours. MEFs were transduced using 5 ml of filtered supernatant added to 5 ml cell culture medium in the presence of 8 mg/ml polybrene (Sigma-Aldrich). After 24 hours virus was removed. 14 days selection was started 72 hours after infection using 400μg/ml neomycin (Sigma-Aldrich).

In a second infection step, Large T-Antigen-immortalized MEFs were transduced with the Cre-pBabe-puro retrovirus for the Cre-mediated deletion of ATM. Selection was started 72 hours after transduction with 2 μg/ml puromycin for 7 days. Medium was exchanged every Deletion of ATM was verified by PCR of genomic DNA using primer 1 (5' ATC AAA TGT AAA GGC GGC TTC 3'), primer 2 (5'CAT CCT TTA ATG TGC CTC CCT TCG CC 3') and primer 3 (5'GCC CAT CCC GTC CAC AAT ATC TCT GC 3') giving rise to three fragments representing absence or presence of loxP sites or deletion of ATM.

Plasmid construction, generation of recombinant retroviruses and cell line establishment

SNEV cDNA was amplified by PCR and ligated into the retroviral plasmid pLenti6. The SNEV S149A point mutant was generated using the QuikChange Multi Site-Directed Mutagenesis Kit (Agilent, Santa Clara, CA, USA). The negative control vector contained only the blasticidin resistance gene. Retroviral particles were generated according to the manufacturer's protocol (Invitrogen). For cell line establishment, HUVECs (PD14) were infected with retroviral particles according to the manufacturer's protocol (Invitrogen, now life technologies, Grand Island, NY, USA) at a multiplicity of infection (MOI) of 4 in EBM with 10% FCS supplemented with 8 μg/ml Polybrene. Thereafter, transfectants were selected using 5 μg/ml Blasticidin. Arising cell clones (similar number in all experiments) were grown as mass culture. PDs post transfection (pT) were calculated starting with the first passage after selection was completed.

Antibodies

Prp19/Pso4 rabbit polyclonal antibody was from Bethyl Laboratories (Montgomery, TX, USA) #A300-102A. ATM antibody [2C1] mouse monoclonal antibody was from Gene Tex (Irvina, CA, USA) #GTX70103. anti-pSNEV(S149) rabbit polyclonal antibody was generated by Moravian Biotechnology Ltd. (Bratislava, Slovakia). Gamma H2A.X (phospho S139) mouse monoclonal antibody [9F3] was from Abcam (Cambridge, UK) #26350. β-Actin mouse monoclonal antibody was from Sigma-Aldrich (St.Louis, MO, USA) #A-5441. GAPDH rabbit antibody FL-335 was from Santa Cruz (Santa Cruz, CA, USA) #sc-25778.

SDS PAGE and Western Blotting

For SDS PAGE, protein samples were mixed with 4x SDS loading dye (240 mTris-Cl pH 6.8, 8% SDS, 40% glycerol, 0.05% bromophenol blue, 5% β-Mercaptoethanol) and heated to 75°C for 10min. Samples were separated on a NuPAGE 4-12% Bis/Tris polyacrylamide gel (Invitrogen, Carlsbad, CA; USA) in MOPS buffer at 200V. Electrophoresis and blotting to PVDF membrane (Roth, Karlsruhe, Germany) were performed using the XCell SureLock®Mini-Cell (Invitrogen, now life technologies, Grand Island, NY, USA) in accordance to the manufacturer's protocol. After incubating with blocking buffer (3% skim milk powder in PBS with 0.1% Tween-20) on an orbital for 1h at room temperature or overnight at 4°C, the membranes were incubated for 1h or overnight with the primary antibody diluted in blocking buffer, followed by 1h incubation with secondary anti-Rabbit-IR-Dye 800 and/or anti-Mouse-Alexa 680 (Licor), both diluted 1:10000 in blocking buffer. Both antibody incubations were followed by 3 washes with PBS with 0.1% Tween-20. Membranes were scanned using the Odyssey infrared imaging system (LI-COR, Lincoln, NE, USA). Mouse anti-?Actin (Sigma-Aldrich, St.Louis, MO, USA) or rabbit anti-GAPDH (Santa Cruz, Santa Cruz, CA, USA) were used as loading controls.

Dephosphorylation

For dephosphorylation experi-ments, H₂O₂ treated or untreated HeLa cells were harvested by trypsinization, collected by centrifugation at 170g for 10 min, washed twice with PBS and resuspended in NEB3 Buffer (New England Biolabs, Ipswich, MA, USA), in which Calf intestinal phosphatase (CIP; New England Biolabs, Ipswich, MA, USA) has optimal activity. Cells were lysed by sonication and subsequent centrifugation at maximum speed for 30 min at 4°C. Cleared lysates were split and one half was incubated with 10U CIP for 30 min at 37°C, whereas the other half was mock treated. Subsequently, samples were submitted to Western Blot analysis.

Anti-GFP Trap IP

GFP-SNEV-Hela were grown to 90% confluence in T175 roux flask and incubated with 100 μM H₂O₂ to induce SNEV phosphorylation or left untreated as negative control. Cells were scraped on ice in 3ml Lysis buffer (50 mM Hepes-KOH pH 7.5, 5 mM EDTA pH8, 150 mM KCl, 10% glycerol, 1% Triton X-100, 20 mM β-glycerophosphate, 10mM Na-pyrophosphate, 10 mM NaF, 1 mM DTT, 1 mM Na₃VO₄, 0.1mM PMSF, 20 μg/ml Leupeptin, 20 μg/ml Chymostatin, 20 μg/ml Pepstatin, 1μM Okadaic acid) per T175 and sonicated using 3 bursts of 10s. Lysate was cleared by centrifugation for 30 min at full speed. All centrifugation stepy were carried out at 4°C. 30 μl GFP-Trap Beads (ChromoTek, Munich, Germany) were equilibrated by washing thrice in 500 μl ice cold wash buffer (50mM Hepes-KOH pH 7.5, 5mM EDTA pH8, 150mM KCl, 10% glycerol, 0.05% NP-40, 20 mM β-glycerophosphate, 10 mM Na-pyrophosphate, 10mM NaF, 20 μg/ml Leupeptin, 20 μg/ml Chymostatin, 20 μg/ml Pepstatin) and collected by centrifugation at 2,700 x g for 2 min. 1 ml Lysate corresponding to 1 mg total protein was added to the beads mixed on an overhead shaker for 1h at 4°C. Beads were collected by centrifugation for 2 minutes at 2,000 x g. Discard supernatant and wash beads 5x times with 500μL ice cold wash buffer (ocadaic acid added only for the first two washing steps) and finally resuspended in washing buffer.

Mass spectrometric analysis

Beads were washed five times with ammonium bicarbonate buffer (50mM ABC). Disulfide bonds were reduced by incubation with dithiothreitol (DTT, 5% w/w of the estimated amount of protein) for 30 min at 56°C and Cys-residues were subsequently alkylated with iododacetamide (IAA, 25% w/w of the estimated amount of protein) for 20 min at RT in the dark. DTT (2.5% w/w of the estimated amount of protein) was added to consume excess IAA and proteins were digested with subtilisin for 1 hour at 37°C. Digests were stopped by addition of trifluoro acetic acid (TFA) to approx. pH 3. 10% of the peptide mixture was analysed directly, the remaining 90% were enriched for phosphopeptides by the use of TiO2 as described in [48]. Peptides were separated on a U3000-HPLC-system (Dionex, Sunnyvale, CA, USA). Peptides were loaded on to a the trapping column (PepMAP C18, 0.3 × 5 mm, Dionex) with 0.1% TFA as loading solvent and then eluted onto an analytical column (PepMAP C18, 75 μm × 150 mm, Dionex) with a flow rate of 300 nl/min and a gradient from 0% solvent B to 100% solvent B in 90min, followed by a washing step of 10 min with 10% solvent B and 90% solvent C (solvent B: 40% acetonitril (ANC), 0.08% formic acid, solvent C: 80% ACN, 10% trifluorethanol, 0.08% formic acid). The HPLC system is online coupled to a Velos Orbitrap mass spectrometer (Thermo Scientific, Waltham, MA, USA) equipped with an ESI-source (Proxeon, now Thermo Scientific). A full scan (scan range 400-1800 Th, resolution 60.000) was followed by MS2 analysis by collision-induced dissociation (CID) of the 20 most intense precursors in the linear iontrap. Normalized collision energy was set to 35%, activation Q at 25 and activation time at 30 ms. Peptide identification was performed using the SEQUEST algorithm in the ProteomeDiscoverer 1.3.0.339 software package (Thermo Scientific). Carbamidomethylation of Cys was set as static modifications, phosphorylation of Ser/Thr/Tyr and oxidation of Met were set as variable modifications. Spectra were searched against a small database plus contaminants for a fast PTM analysis. Search parameters were no protease specificity, a peptide tolerance of 2 ppm, a fragment ions tolerance of 0.8 Da. The results were filtered at the XCorr values to an FDR of 1% on the peptide level. The probability of phosphosite localization was calculated using the phosphoRS 2.0 software [49] implemented into the Proteome Discoverer.

Indirect immunofuorescence staining

Cells were seeded on coverslips one day prior to immunofluorescence staining. The next day, cells were washed with PBS and fixed with 3.7% (w/v) for 20 min at room temperature. Permeabilizationwas performed with 0.1% Triton X-100 in PBS for 10 min at room temperature. Cells were incubated with primary antibodies diluted in PBS with10% FCS for 1 h, washed 3 times for 10 min withPBS, incubated for 1 h with the appropriate secondaryantibodies diluted in PBS with 10% FCS, and washed 3 timesfor 10 min with PBS. Anti-pSNEV (S149) was diluted 1:250. As secondary antibodies, anti mouse dyelight488 1:1000 and dyelight649 1:1000anti-rabbit or anti-mouse antibodies were used. Microscope as described previously.

To visualize the nuclei, cells were counterstained with 4',6-diamidino-2-phenylindole (DAPI). Cells were washed and fixed as above and subsequently incubated with 100ng/ml DAPI in PBS for 10 min at room temperature.

After staining, slides were mounted on cover slips using slow fade gold Mounting Medium for Fluorescence (life technologie, Grand Island, NY, USA) and sealed with nail polish.

Microscopyand image analysis were carried out using a Leica SP5 II laser scanning confocal microscope (Leica Microsystems CMS, Mannheim, Germany).

Comet assay

Single-and double-stranded DNA damage in SNEVwt and SNEV S149A overexpressing HUVEC under basal conditions was measured by single-cell gel electrophoresis (comet assay) under alkaline conditions as described previously [50]. At least 100 randomly selected cells per slide were examined for the presence or absence of comets using fluorescence microscopy. The cells were assigned to five different categories according to their DNA content in tail using the TriTek CometScore software. Percentage of cells in each category was calculated.

Apoptosis staining

Cells were detached using 0.1% Trypsin/0,02%EDTA and stained with Annexin V-Bacific Blue (Invitrogen, now life technologies) and PI (Roche, Basel, Switzerland) according to the manufacturer's instructions. Analysis of the percentages of apoptotic and necrotic/late-apoptotic cells was performed using a BD FACS Canto II flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA) and the FCS Express V3 software (De Novo Software, Los Angeles, CA, USA).