Alessandra Pagano, Claire Pitteloud, Coralie Reverdin, Isabelle Me´trailler-Ruchonnet, Yves Donati, and Constance Barazzone Argiroffo
Departments of Pediatrics and Pathology-Immunology, University of Geneva, Medical School, Geneva, Switzerland
Hyperoxia induces extensive DNA damage and lung cell death by apoptotic and nonapoptotic pathways. We analyzed the regulation of Poly(ADP-ribose)polymerase-1 (PARP-1), a nuclear enzyme acti-vated by DNA damage, and its relation to cell death during hyper-oxiain vitroandin vivo.In lung epithelial-derived A549 cells, which are known to die by necrosis when exposed to oxygen, a minimal amount of PARP-1 was cleaved, correlating with the absence of active caspase-3. Conversely, in primary lung fibroblasts, which die mainly by apoptosis, the complete cleavage of PARP-1 was concomi-tant to the induction of active caspase-3, as assessed by Western blot and caspase activity. Blockade of caspase activity by Z-VAD reduced the amount of cleaved PARP-1 in fibroblasts. Hyperoxia induced PARP activity in both cell types, as revealed by poly-ADP-ribose accumulation. In A549 cells, the final outcome of necrosis was dependent on PARP activity because it was prevented by the PARP inhibitor 3-aminobenzamide. In contrast, apoptosis of lung fibroblasts was not sensitive to 3-aminobenzamide and was not affected by PARP-1 deletion.In vivo, despite evidence of PARP activa-tion in hyperoxia-exposed mouse lungs, absence of PARP-1 did not change the extent of lung damage, arguing for redundant oxidative stress–induced cell death pathways.
Keywords:apoptosis; caspase-3; hyperoxia; necrosis; PARP-1 High oxygen exposure has been used as a valuable model for acute respiratory distress syndrome (ARDS) in rodents and is characterized by extensive parenchymal cell death (1). The un-derstanding of the molecular mechanisms and signaling pathways leading to alveolar cell death is essential for the development of efficient therapeutic treatments to protect the alveolo–capillary barrier. We previously showed that mouse alveolar cells exposed to 100% oxygen exhibit apoptotic and necrotic features in vivo.
However, it is difficult to define which cell type dies by apoptosis or necrosis exclusively according to morphologic criteria (2). Nu-merous in vitro studies indicate that the mechanism of hyperoxia-induced cell death depends on the cell type–specific response.
Indeed, human (A549) and mouse (MLE12 and MLE15) epithelial cell lines exposed to hyperoxia show several character-istics of oncotic/necrotic cell death besides the activation of an early apoptotic signaling pathway (3–6). Murine macrophages and fibroblast cell lines (Rat1) seem to undergo apoptosis be-cause they show several apoptotic features, including DNA lad-dering and caspase activation (7, 8). Recently, several reports
(Received in original form November 17, 2004 and in final form September 1, 2005) This work was supported by the Swiss National Research Foundation #3200-067865.02, and by the Wo¨lfermann-Naegele, Novartis, and Lancardis Foundations.
Correspondence and requests for reprints should be addressed to Dr. Constance Barazzone Argiroffo, Departments of Pediatrics and Pathology, Centre Me´dical Universitaire, 1, rue Michel Servet, 1211, Geneva, 4, Switzerland. E-mail: constance.
barazzone@hcuge.ch
Am J Respir Cell Mol Biol Vol 33. pp 555–564, 2005
Originally Published in Press as DOI: 10.1165/rcmb.2004-0361OC on September 8, 2005 Internet address: www.atsjournals.org
have suggested that intermediate patterns of cell death may exist (4, 5).
Hyperoxia generates reactive oxygen species, leading to mas-sive oxidation and DNA damage (9). DNA oxidation and strand breaks have been directly detected by comet assay in cultured epithelial cell lines exposed to oxygen (10, 11) and in type II alveolar epithelial cells isolated from hyperoxia-exposed mice (12). In vivo, although free radical–mediated DNA strand breaks have not been directly shown in the lung, DNA damage has been revealed by terminal transferase nick end-labeling (TUNEL) and DNA electrophoresis (DNA laddering) in oxygen-exposed mouse lungs (2, 13).
Poly(ADP-ribose) polymerase-1 (PARP-1) is the most abun-dant nuclear enzyme of the PARP family that is activated in response to DNA damage and participates in DNA repair, geno-mic integrity, and cell death (14). PARP-1 binds rapidly to DNA strand breaks and adds branched poly(ADP-ribose) (PAR) poly-mers, using nicotinamide adenine dinucleotide (NAD�) as a substrate, on itself and other nuclear proteins (e.g., histones), thereby facilitating the action of DNA repair enzymes (15).
PARP-1 has been described to be involved in the regulation of cell death (14). The presence of cleaved PARP-1 has been considered a characteristic hallmark of apoptosis. Caspases, in particular caspase-3, are known to cleave PARP-1 and therefore inhibit its activity (16, 17). On the other hand, excessive DNA damage induces massive PARP activation, leading to the deple-tion of cellular stores of NAD�and ATP and consequent energy failure followed by necrotic cell death (suicide theory) (18).
PARP activation has been reported in several patho-physiologic conditions characterized by oxidative stress, cell death, and inflammation, such as hemorrhagic shock, cerebral ischemia, asthma, and lipopolysaccharide-induced acute lung injury. The specific deletion of the PARP-1 gene is associated with a benefi-cial effect and decreased cell death in those models (19–22).
NAD�depletion and PAR accumulation have been described in total rat lungs exposed to oxygen (23), suggesting a role for PARP in response to the oxygen-mediated oxidative stress.
In this context, we hypothesized that during acute exposure to oxygen, PARP-1 might participate in directing the cell death response. We studied PARP-1 regulation in two pulmonary cell populations. We chose the A549 epithelial cell line, known to die by necrosis (3, 4), and we isolated primary fibroblasts from mouse lung that died primarily by apoptosis during hyperoxia.
We also studied the role of PARP-1 in vivo by using PARP-1
�/� mice and by treating animals with the PARP inhibitor 3-aminobenzamide (3-AB) during hyperoxia. In vitro, PARP-1 was differentially regulated according to the cell type and the mode of cell death. In particular, necrosis in epithelial A549 cells but not apoptosis in lung fibroblasts was dependent on PARP activity. Despite the induction of PARP activity, which may contribute to the cellular energy failure, the absence of PARP-1 was not sufficient to prevent hyperoxia-induced cell death in vivo.
Chapter 2
556 AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL 33 2005
MATERIALS AND METHODS Cell Culture and Hyperoxia Exposure
Human lung adenocarcinoma A549 cell line (ATCC CCL185) were grown in F12K medium (Kaighn’s modification; Gibco, Paisley, UK), supplemented with 1% penicillin-streptomycin (Gibco) and 10% fetal calf serum. Primary mouse lung fibroblasts were isolated from lung explants of Sv129 PARP-1�/�or Sv129 PARP-1�/�mice as pre-viously described (4) and grown in Dulbecco’s modified Eagle’s medium (glucose 1,000 mg/l; Sigma-Aldrich, Buchs, Switzerland) supplemented with 1% penicillin-streptomycin and 10% fetal calf serum. Cell plates were kept in an incubator at 37�C and monitored daily until fibroblasts reached confluence. Within 2 wk of culture initiation, cells presented similar spindle-shape morphology, as assessed by optical microscopy (24, 25). Approximately 90% of the cells were positive for the surface marker Thy1.2 and negative for the hematopoietic marker CD45 when analyzed by flow-cytometry. Lung fibroblasts were positive for vimentin, as detected by immunofluorescence (not shown). Cell phenotype was analyzed every second passage, and fibroblasts were used for the experi-ments between passages 6 and 12.
For each experiment, cells were plated at subconfluence (70%) and allowed to adhere 24 h before the experiment. Hyperoxic conditions were achieved by placing the cells in sealed glass chambers filled with 95% O2–5% CO2at 37�C for up to 96 h. The oxygen concentration was checked at the beginning and end of the exposure period by an oxygen analyzer (OM 11; Beckman, Fullerton, CA) as previously described (26). Control cells were kept in air (21% O2–5% CO2) at 37�C. Medium and gases were replaced daily.
3-AB (Alexis Biochemicals, San Diego, CA) was added to cell cultures at the final concentration of 3 mM because at this dosage the pharmacologic inhibitor presented minimal toxicity and was effective in hyperoxia. The pan-caspase inhibitor Z-VAD-fmk (Caltag Labora-tories, Burlingame, CA) was added at the final concentration of 100�M as described (27).
Determination of Cell Death
Assessment of intracellular lactate dehydrogenase release.Lactate dehy-drogenase (LDH) release, a marker of cell death, was measured in cell culture supernatants: 200�l of culture medium was removed, centri-fuged for 5 min at 400 g, and stored at 4�C. For intracellular LDH determination, cells were lysed by adding 200 �l of fresh medium containing 7.5% Triton X-100 (final concentration 1%). LDH extra- and intracellular content was measured using a colorimetric assay (Roche Molecular Biochemicals, Rotkreuz, Switzerland). LDH release was cal-culated as the ratio of extracellular to total (extracellular�intracellular) LDH content (mean of triplicates�SD). Values were expressed as the percentage of total releasable LDH.
Trypan blue dye exclusion. Nonadherent and adherent cells were collected and stained with an equal volume of 10% Trypan blue dye solution. The extent of cell death was expressed as the percentage of blue (dead, floating�adherent) cells over total cell number.
Analysis of nuclear morphology. Nuclear morphology was assessed by Hoechst 33258 (Sigma-Aldrich) staining. Cells were plated on cov-erslips placed in 30 mm Petri dishes. Adherent cells were incubated with Hoechst 33258 (2�g/ml) in PBS for 5 min at 37�C, washed, and fixed with 4% buffered formalin or with 50% acetone/methanol. Slides were mounted with Mowiol 4–88 (Sigma-Aldrich) and visualized with a fluorescence microscope (Zeiss Axiophot 1 equipped with an Axiocam color CCD camera; Carl Zeiss, Oberkochen, Germany). Nuclei were scored as apoptotic if they appeared smaller and brighter, indicating nuclear condensation. Multiple high-power (40�) fields were averaged, and results were expressed as the percentage of total cells counted (around 600 cells/sample) in three independent experiments.
Quantification of apoptosis by flow cytometry. Cells were washed in PBS (1�), incubated with Annexin-V-FITC according to the manufac-turer’s instruction (BD Biosciences-Pharmingen, Heidelberg, Germany), and stained with propidium iodide (PI) (Sigma Aldrich). The analysis was performed with a FACScan flow-cytometer (BD Biosciences-Pharmingen).
Caspase-3 Activity
Cells (1 � 106/condition) were collected, washed in PBS (1�), and resuspended in hypotonic buffer containing 25 mM HEPES, 5 mM MgCl2, 1 mM aprotinin, 1 mM EDTA, and 1 mM Pefabloc. Protein content was measured using a BCA protein assay kit (Pierce, Rockford, IL). For the caspase activity measurement, 20–60�g of proteins sus-pended in 20�l of the buffer were distributed in a black microclear bottom 96-well plates (Greiner Bio-One GmbH; Frickenhausen, Germany). A mix (180�l) containing 10 mM Hepes, 0.1% dimethyl-ammonio-l propanesulfonate, 1% saccharose, 5 mM DTT, and 30�M of the specific substrate for caspase3 coupled to a fluorochrome (DEVD-AFC, stock at 12.5 mM in DMSO) was added to the sample.
The accumulation of the fluorescent product (360 nm excitation/ 530 nm emission) was recorded during 3 h using a benchtop scanning fluo-rometer (FlexStation II; Molecular Devices, Bucher Biotec AG, Basel, Switzerland). Analysis of the data was performed using SoftMax Pro and Excel software.
Immunocytochemistry
Cells were fixed in trichloroacetic acid 10% and incubated with an anti-PAR mouse monoclonal antibody (H10, ref. no. 804–220-R100; Alexis Biochemicals) (dilution: 1/100) overnight at 4�C in PBS-Tween 0.05%
as previously described (28, 29). After washing cells in PBS-Tween 0.1%, a biotinylated antimouse IgG, Fc-specific, antibody (ref. no. 115–
065–071; Jackson Laboratories, San Diego, CA) (dilution: 1/500) was added and labeled with streptavidin-FITC (Caltag) (dilution: 1/500).
4,6-Diamidino-2-phenylindole (DAPI) (Roche Diagnostics, Rotkreuz, Switzerland) (dilution 1/200) was added to visualize nuclei. Slides were mounted and analyzed as described previously.
Animals and Hyperoxia Exposure
PARP-1�/�(Sv129 mice genetic background) and their littermates (�/�) were furnished by Dr. Wang’s laboratory (CIRC, Lyon, France) (30) and bred in our animal facility. Mice were identified by PCR according to the conditions described by Wang (31). PARP-1 protein deletion was verified by Western blot.
Experiments were performed with 8- to 10-wk-old mice weighing 20–25 g. PARP-1�/�and PARP-1�/�mice were placed in a sealed plexiglas chamber and exposed to air or 100% O2for 72 h as described (32). In some experiments, C57BL/6 mice were treated daily with the PARP inhibitor 3-AB (20 mg/kg/d dissolved in DMSO/PBS and given intraperitoneally) or the vehicle (DMSO/PBS) during oxygen exposure.
After mice were killed, lung damage was evaluated macroscopically, and a score was given (33). Pulmonary edema was determined by measuring wet lung weight (34). The right lung was immediately frozen in liquid nitrogen and stored at –80�C for further analysis. In some experiments, the left lung was fixed and stored for histologic analysis.
All study protocols were approved by the local ethics committee on animal experiments (Office Ve´te´rinaire Cantonal of Geneva).
Immunohistochemistry
Left lungs were fixed by intratracheal instillation of 4% buffered forma-lin in phosphate buffer and embedded in paraffin. Paraffin-embedded lung tissue sections (3�m) were stained with hematoxylin and eosin for histologic evaluation or processed for immunohistochemistry ac-cording to previously described protocols (33). Briefly, sections were deparaffinized, rehydrated, and cooked 3�5 min in a microwave oven to facilitate the access of the antibody. After cooling, samples were blocked with 5% BSA in Tris-buffered saline and incubated overnight with an anti-PAR monoclonal antibody (clone H10; Alexis) (dilution:
1/200). As secondary antibody, a biotinylated anti-mouse IgG, Fc-specific, antibody (Jackson) (dilution: 1/2,000) was added and labeled with streptavidin-biotin-peroxidase complex (A�B kit; Vector Labora-tories, Burlingame, CA). Sections were counterstained with Hemalum.
The peroxidase activity was revealed with the addition of the substrate 3,3�-diaminobenzidine tetrahydrochloride (Sigma-Aldrich). Normal mouse IgG was used instead of the primary antibody as negative control for nonspecific binding as previously described (33).
Western Blot Analysis for Cellular and Lung Extracts
Cells were washed, trypsinized, and collected by centrifugation at 400 �g. The cell pellet was lysed in total lysis buffer (50 mM Tris
Pagano, Pitteloud, Reverdin,et al.: PARP-1 and Hyperoxia-Induced Cell Death 557 [pH 8.0], 150 mM NaCl, 0.1% SDS, 1% Triton x-100, 0.5%
Na-deoxycolate, proteases inhibitors). A portion of mouse frozen lung was homogenized in the same lysis buffer (40 mg/ml) in a glass Dounce homogenizer with a Teflon pestle. Homogenates were centrifuged (13,000�gfor 15 min at 4�C), and the supernatant was collected for analysis. Samples were analyzed for protein concentration with a Bio-Rad DC Protein assay kit (ref 500–0111, Bio-Bio-Rad Laboratories, Her-cules, CA). Fifty micrograms of total protein cell extracts or 100�g of total protein lung extracts underwent electrophoresis on SDS-polyacryl-amide gels and were blotted to nitrocellulose membranes (Amersham International, Amersham, UK). Membranes were blocked overnight in TBS-T buffer (Tris 0.2 M [pH 7.6], NaCl 1.5 M, 0.1% Tween-20) and 5% milk and incubated with the following antibodies: mouse mono-clonal anti–PARP-1 antibody (C210, ref. no. 556362; BD Biosciences-Pharmingen; dilution 1:500), which recognizes native and cleaved forms of PARP-1; rabbit polyclonal anti-cleaved PARP-1 (Cell Signaling Technology, Beverly, MA) (dilution 1:1000); rabbit polyclonal anti-cleaved caspase-3 (9661-S; Cell Signaling Technology, Beverly, MA) (1:1,000); and rabbit polyclonal anti-actin antibody (AL-20, gift from Gabbiani laboratory) (1:2,000). Native and cleaved PARP-1 were re-vealed on the same membranes. Horseradish peroxidase-conjugated anti-mouse and anti-rabbit (1:3,000 dilution, Bio-Rad Laboratories) were used as secondary antibodies. Bands were visualized with a
chemi-Figure 1. Time-course of cell death in hyperoxia-exposed A549 cells and lung fibroblasts. Cells were exposed to air or hyperoxia for 96 h. LDH release from A549 (A) and fibroblasts (C) was measured and expressed as the percentage of the ratio of extracellular (supernatant, SPN) to total (extracellular�intracellular) LDH content. Values are expressed as mean�SD of triplicates (one representative experiment is shown), ***P� 0.001, 48–72–96 h Hox versus air.B(A549) andD(fibroblasts) are representative fields of cells exposed to air or hyperoxia (72 h) and stained with Hoechst 33258 (magnification 40�). (E) Quantification of apoptotic nuclei of hyperoxia-exposed fibroblasts from three independent experi-ments. Values are expressed as mean�SD. ***P�0.001, 72–96 h Hox versus air. (F) Cells were exposed to air or hyperoxia for 72 h, labeled with Annexin-V-FITC and PI, and analyzed by flow cytometry.
luminescent substrate (ECL; Amersham International). Anti-actin anti-body was used as a control for total protein loading. Quantification of signal intensity was performed on subsaturated films with the Im-agequant software. Values corresponding to native and cleaved PARP-1 at the different time points during hyperoxia were expressed as fold increases over air-exposed control set as 1.
Statistical Analysis
For all parameters measured, the values for all animals in different groups were averaged, and the SD of the mean was calculated. For in vitrostudies, all measurements were performed in triplicate or in duplicate when indicated. The results were expressed as mean values
�SD. The significance of differences between the values of the groups was determined with an unpaired-Student’s t test. Where appropriate, two-way ANOVA with Bonferroni post-tests was used. Significance levels were set at P�0.05.
RESULTS
Hyperoxia Induces a Differential Cell Death Response in Lung Epithelial Cells and Fibroblasts
To determine whether hyperoxia influences PARP expres-sion and activity according to the mode of cell death, we first
Chapter 2
558 AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL 33 2005
characterized cell death in pulmonary epithelial A549 cells and in primary mouse lung fibroblasts. These two cell types were exposed to hyperoxia, and their mortality was compared by measuring LDH release. In A549 cells, there was a significant increase of LDH release starting from 48 h of exposure (P� 0.001) (Figure 1A). Primary fibroblasts (Figure 1C) showed a comparable increase in LDH release. Similar results were ob-tained by counting Trypan blue–positive cells (A549: 60�7.6%
in hyperoxia-exposed cells for 72 h versus 2.3� 2.1% in air-exposed cells; lung fibroblasts: 22.5�0.4% in hyperoxia-exposed cells for 72 h versus 1.3�0.9% in air-exposed cells, n�3, P� 0.001). We then analyzed their morphologic changes by Hoechst 33258 staining. At 72 h of oxygen exposure, A549 cells did not show any increase in fluorescence staining, whereas nuclei appeared swollen compared with air-exposed cells (Figure 1B, compare right versus left). In contrast, several nuclei of primary lung fibroblasts were shrunken and brighter, characteristic signs of apoptosis (Figure 1D, compare right versus left). Quantifica-tion of the condensed nuclei (apoptotic index) in three different experiments showed a significant increase in apoptotic fibro-blasts at 72 h of hyperoxia (P�0.001) (Figure 1E). To confirm these results, we labeled lung fibroblasts with Annexin-V-FITC, a marker of early apoptosis, and with PI, a marker of necrosis, and analyzed them by flow cytometry. At 72 h of oxygen expo-sure, 24.7% of the cells were single positive for Annexin (early apoptotic), and 25.7% were double positive for Annexin and PI (late apoptotic or necrotic), confirming our morphologic obser-vations (Figure 1F).
We next analyzed whether the different morphologic features observed in these two cell types correlated with a differential regulation of caspase-3 activation. It is generally assumed that the morphologic changes observed during apoptosis are mainly dependent on active caspase-3 (35, 36). No cleaved (active form) caspase-3 was detected by Western blot at any time of oxygen exposure in A549 cells (Figure 2A). Conversely, active caspase-3 was present in hyperoxia-exposed primary lung fibroblasts as early as from 48 h of oxygen exposure (Figure 2B, upper panel).
To confirm the presence of active caspase-3 in primary lung fibroblasts during hyperoxia, we measured caspase-3 activity in total cell lysates (Figure 2B, lower panel). Low levels of caspase-3 activity were detected in air-exposed cells and after 24 h of hyperoxia. Caspase-3 activity was strongly induced after 48 h of oxygen exposure, correlating with the detection of active caspase-3 by Western blot. These results show that lung epithelial cells respond to hyperoxia by oncosis/necrosis without caspase-3 activation, whereas lung fibroblasts undergo classical apoptosis.
Differential Regulation of PARP-1 Expression in Pulmonary Cells during Hyperoxia
Because the presence of cleaved PARP-1 is considered a charac-teristic hallmark of apoptosis, we examined the expression of native and cleaved PARP-1 in hyperoxia-exposed A549 and lung fibroblasts. In A549 cells, native PARP-1 was strongly and stably expressed during hyperoxia (Figure 3A, upper panel). Only a small amount of the cleavage product of PARP-1 (89 kD) appeared at 48 h and 72 h of oxygen exposure and was no longer detected at 96 h. Densitometric analysis of three different experiments (Figure 3A, lower panel) confirmed that, apart from some increase of cleaved PARP-1 (P�0.05) at 48–72 h, most PARP-1 remained uncleaved. In primary lung fibroblasts (Figure 3B, upper panel), the cleaved form of PARP-1 appeared concom-itantly with the disappearance of the native form, demonstrating that a complete cleavage of the protein was induced in those cells. The densitometric analysis (Figure 3B, lower panel) showed a significant reduction of the native form of PARP-1 over time
Figure 2. Detection of active caspase-3 in A549 cells and lung fibro-blasts. Protein extracts (50�g) from A549 (A) and mouse lung primary fibroblasts (B, upper panel) exposed to air or hyperoxia at different times were analyzed by Western blot analysis with anti-cleaved, caspase-3–specific antibody. Results represent three different experiments. Dex:
Thymus extract (15�g of protein) from a mouse treated with 250�g of dexamethasone (intraperitoneally) and killed 6 h later was used as a positive control. Caspase-3 activity (B,lower panel) was measured in total cell lysates of lung fibroblasts exposed to air or hyperoxia at the
Thymus extract (15�g of protein) from a mouse treated with 250�g of dexamethasone (intraperitoneally) and killed 6 h later was used as a positive control. Caspase-3 activity (B,lower panel) was measured in total cell lysates of lung fibroblasts exposed to air or hyperoxia at the