当前位置: 首页 > 医学版 > 期刊论文 > 内科学 > 内分泌学杂志 > 2005年 > 第3期 > 正文
编号:11168249
The Role of Poly (Adenosine 5'-Diphosphate-Ribose) Polymerase in the Response of Pituitary Tumor Cells to Reactive Oxygen Species
     Max-Plank-Institute of Psychiatry, Neuroendocrinology Group, D-80804 Munich, Germany

    Address all correspondence and requests for reprints to: Dr. Ulrich Renner, Max-Planck-Institute of Psychiatry, Neuroendocrinology Group, Kraepelinstr. 10, D-80804 Munich Germany. E-mail: renner@mpipsykl.mpg.de.

    Abstract

    As an enzyme implicated in the stress response, we investigated poly(ADP-ribose) polymerase (PARP) in the response of GH3 rat pituitary tumor cells to oxidants. These cells are unusual in that they undergo rapid cell death (90 min) with low doses of the prooxidant, H2O2 (50–200 μM), whereas at higher doses (1 mM), death occurs some hours later (4–5 h). Measurement of PARP activity shows that low doses of H2O2 (50–200 μM) fail to increase the activity of PARP, whereas at 0.5 and 1 mM, the enzyme becomes activated. In parallel with the activation of PARP, cellular ATP concentrations fall at high H2O2 doses and the PARP inhibitors, 3-aminobenzamide and nicotinamide (NIC) partially prevent this fall. Using NIC to inhibit PARP activity, we show that treatment of cells with NIC before the addition of H2O2 (0.5–1 mM), results in rapid cell death (90 min). In contrast, prior exposure to H2O2 (0.5–1 mM) for 1 h, before withdrawal and exposure to 1 mM NIC, allows cell survival for many hours. These data suggest that PARP is involved in blocking rapid death of GH3 cells in response to oxidants. In contrast to other cell types tested here, in which inhibitor studies show that PARP is activated at low H2O2 doses and this decreases the extent of apoptosis, GH3 cells are unable to sufficiently activate PARP to prevent rapid cell death.

    Introduction

    THE ENZYME, POLY(ADP-ribose) polymerase (PARP) is a chromatin bound enzyme that catalyzes the poly(ADP-ribosyl)ation of histone and nonhistone proteins using the nucleotide, nicotinamide adenine dinucleotide (oxidized form) (NAD+), as substrate (1). The primary cellular function of this enzyme is not entirely certain; however, it becomes active in response to DNA strand breakage and therefore is likely to have a role in DNA repair (2). More recently PARP has come under scrutiny as a substrate for caspase-3 (3), an enzyme activated during apoptotic cell death. Caspase-3 is activated by proapoptotic stimuli, and it cleaves the 116-kDa PARP protein into 90- and 26-kDa fragments that contain the active site and DNA binding domain, respectively. The significance of this may lie in the observation that a Ca2+-, Mg2+-dependent endonuclease, involved in cleaving DNA during apoptosis, is a substrate for PARP and it is maintained in an inactive form by poly(ADP-ribosyl)ation (4).

    Reactive oxygen species (ROS) are a major class of DNA-damaging agents. These small and highly reactive molecules are produced within cells as a consequence of the incomplete reduction of molecular oxygen, principally in mitochondria (5). They are also generated by chemotherapeutic agents (6) and in response to radiotherapy (7). Research over the past decade has provided ample evidence that these agents induce apoptotic cell death. Studies suggest that ROS have a role in the way the tumor suppressor gene, p53, induces apoptosis (8). Others have demonstrated that changes in mitochondrial function, one of the cardinal features of apoptosis, are brought about by ROS (9). After cellular ROS exposure, studies demonstrated the activation of PARP (10, 11, 12, 13). Studies showing that the PARP inhibitor, nicotinamide, enhances radiation-induced cell death (14) suggest that PARP activity suppresses apoptosis. Other studies have shown that the same compounds block apoptosis in neuronal cell systems (15). One of the consequences of strong PARP activation is the depletion of cellular NAD+ concentrations due to heavy NAD+ use and the block that this imposes on glycolytic and mitochondrial ATP generation (16). Substantial and sustained ATP depletion leads to the inhibition of the plasma membrane Na+/K+ ATPase that in turn results in an increase in cytosolic volume and rupture of the cell membrane. By blocking PARP, severe ATP depletion should be avoided, preventing necrotic cell death. This idea has led to the proposal that PARP might form a target for the development of drugs against disorders linked with ROS-induced cell death, such as neurodegeneration and stroke (13).

    In contrast to the role of PARP in the pathophysiology of disease, the primary physiological function of PARP may be as a component of the early stress-signaling mechanism in response to ROS. PARP activation in response to DNA damage may serve to suppress the inappropriate activation of nuclease activity and the onset of nuclear changes, one of the characteristics of apoptosis. In this respect, our studies on the role of the estrogen receptor on the survival of the lactosomatotroph GH3 rat pituitary tumor cells demonstrated a nonclassical response to ROS (17). A peculiarity shown by these cells is that death induced by low concentrations (50–200 μM) of the prooxidant, hydrogen peroxide (H2O2), is induced more rapidly than at substantially higher doses (0.5–1 mM). The characteristics of this form of cell death were also unusual in that extensive nucleosomal DNA fragmentation occurred together with a loss of mitochondrial function and cell membrane integrity. This form of cell death therefore had some characteristics of apoptosis and some of necrosis. In view of reports on the role of PARP in response to ROS, we have determined whether PARP plays a role in the unusual response of GH3 cells to the prooxidant, H2O2. This report describes our studies.

    Materials and Methods

    Reagents

    All reagents for cell culture were obtained form Invitrogen (Karlsruhe, Germany), Falcon (Heidelberg, Germany), Nunc (Wiesbaden, Germany), Seromed (Berlin, Germany), Flow (Meckenheim, Germany), and Sigma Chemicals (St. Louis, MO). The cell death ELISA was obtained from Roche Diagnostics (Mannheim, Germany). Fluorescent probes were obtained from Molecular Probes (Cambridge Bioscience, UK). Tritiated nicotinamide adenine dinucleotide ([3H-]NAD+) was obtained from Amersham BioSciences (Freiburg, Germany). Other reagents were obtained from Sigma-Aldrich (Taufkirchen, Germany).

    Cell culture

    Pituitary GH3 cells as previously described (18) were maintained in DMEM with phenol red, penicillin (50 U/ml), streptomycin (50 μg/ml), amphotericin (2 μg/ml), and 10% fetal calf serum (GIBCO, Gaithersburg, MD). Human umbilical vascular endothelial cells (HUVECs) were prepared as previously described (19) and maintained in DMEM with 20% fetal calf serum. EA.hy 926 endothelial cells and OAW 42 ovarian carcinoma cells were cultured in DMEM, according to our previously published studies (17, 20). For the majority of studies reported here, experiments were conducted on cells seeded into 48-well culture plates at densities over the range 4,000–10,000 cells/cm2. Cells were kept at 37 C in an atmosphere of 5% CO2 in air.

    Microscopy

    Normal light (image modulated contrast) in combination with fluorescence microscopy was performed with a DMIL (Leica, Heidelberg, Germany) fitted with a SPOTJUNIOR digital camera (Diagnostic Instruments, Oxnard, CA). To determine the integrity of cell membranes, a mixture of propidium iodide (PI) and Hoechst 33342 was added to the cell culture medium to give final concentrations of 10 and 1 μg/ml, respectively. After 5 min at 37 C, cells were visualized after exposure to normal and UV light (380 nm) using a wide-band pass filter. Cells with disrupted membranes preferentially gave red nuclear fluorescence due to the uptake of PI. Cells with intact membranes gave blue nuclear fluorescence due to the uptake of the cell permeable Hoechst 33342 fluorochrome.

    Cellular metabolic activity

    An estimate of metabolic activity was made by indirectly determining cellular reducing power using the diazo dye 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT). In the presence of reducing equivalents in the form of nicotinamide adenine dinucleotide hydroxide and nicotinamide adenine dinucleotide phosphate reduced, the yellow formazan dye is reduced to blue insoluble (in aqueous solution) crystals. After the treatment period, MTT is added to the culture supernatant to give a final concentration of 0.1 mg/ml. After a period of 30 min for GH3 cells or 4 h for other cell types, a solution of sodium dodecyl sulfate in HCl is added to give final concentrations of 10% and 10 mM, respectively. After overnight solubilization of the blue crystals, aliquots of the solution are taken for measurement in a multiwell spectrophotometer at 540 nm.

    Mode of cell death

    Lactate dehydrogenase (LDH) activity released into the culture supernatant as a marker of cell membrane damage was used to complement the fluorescence microscopy studies with Hoechst 33342 and PI. After treatment, 100-μl aliquots of medium were mixed with a solution of lactate, NAD+, and tetrazolium salt, as described by the manufacturer of the assay kit, Roche Diagnostics. LDH activity in the supernatant was spectrometrically determined at 540 nm.

    DNA fragmentation was used as a marker for apoptosis. DNA fragmentation is observed in both apoptosis and necrosis, but in the former, the degree of DNA fragmentation is far greater. We previously validated an assay kit made by Roche Diagnostics that is able to discriminate between these two modes of cell death. This assay employs both anti-DNA and antihistone antibodies, and it fails to measure DNA fragmentation that occurs with the endothelial cell line, EA.hy926, when this is exposed to high doses of H2O2 (17). At doses of H2O2 around 1 mM, EA.hy 926 cells show classical signs of necrosis in which membrane integrity is lost.

    PARP activity and ATP concentrations

    PARP enzyme activity was detected by measuring the incorporation of [3H]-NAD+ into acid-insoluble material as previously described (21). In brief, after incubation over 45 min to 5 h with H2O2, 1 x 106 cells were harvested from 25-cm2 tissue culture flasks, washed with PBS and resuspended (100 μl) in ice-cold permeation medium [10 mM Tris-HCl buffer (pH 7.8), 1 mM EDTA, 4 mM MgCl2, and 1 mM ?-mercaptoethanol]. After 15 min on ice, the samples were mixed with 200 μl of assay mixture [50 mM Tris-HCl buffer (pH 8.5), 4 mM MgCl2, and 0.5 mg/ml BSA], and the reaction was started by the addition of 50 μl of 200 μM [3H-]NAD+ (35 Ci/mmol) (Amersham Life Sciences, Buckinghamshire, UK). The samples were incubated for 30 min at 25 C, and the reaction was stopped by the addition of 350 μl of 20% trichloroacetic acid. The acid-insoluble precipitate was collected on glass fiber filters (GF/C, Whatman, Millipore, UK). The filters were washed twice with 7.5% trichloroacetic acid and cold ethanol and dried. Radioactivity was determined using a liquid scintillation counter.

    ATP concentrations were determined by a luminescent assay as described in a previous publication (20).

    Mitochondrial membrane potential (m)

    The m was determined by a nonquantitative technique using the cationic fluorochrome 5,5',6,6'-tetrachloro-1,1,3,3'-tetraethylbenzimidolycarbocyanine iodide (JC-1). Cells were treated with JC-1 at final concentration 1 μg/ml for 10 min before exposure to light at 480 nm and observation using a Leica DMIL fitted with a low noise SPOTJUNIOR digital camera (Diagnostic Instruments) and with a wide-band pass filter. As a cationic dye, JC-1 is taken up by mitochondria according to the membrane potential (22). At low m, the dye exists as a monomer and fluoresces green. At high m, JC-1 forms J-aggregates and fluoresces red.

    Statistical analysis

    Values presented are reproduced as mean ± SD with a minimum of four replicates (for some data points, error bars are contained within symbols and cannot be observed). Analysis of significance for the difference in mean values for any two data sets in an experiment was calculated using the Student’s t test for independent samples. Data presented are representative of experiments conducted on at least five separate occasions, in which both qualitatively and quantitatively similar results were obtained. With the exception of some data (those presented in Fig. 8A), the coefficient of variation for data points did not exceed 20%.

    FIG. 8. Effect of NIC on H2O2-induced LDH release, DNA fragmentation, and change in metabolic activity in HUVECs (A), transformed endothelial cells, EA.hy 926 (B), and the ovarian carcinoma cell line, OAW 42 (C).

    Results

    Characteristics of ROS-induced pituitary tumor cell death

    H2O2 freely diffuses across cell membranes, and it is known that this compound induces the intracellular production of ROS (23). The tumor cell line GH3 was exposed to H2O2 over the dose range 0.05–1 mM for up to 5 h. For seven experiments conducted in this way, the proportion of viable cells was estimated over time by the addition of a mixture of the fluorochromes, PI, and Hoechst 33342. After 60 min in the presence of concentrations of H2O2 over the range 50–200 μM, less than 5% of the cell population was observed to have taken up PI. This increased to 57% (43–68%, 95% confidence interval) after 90 min exposure. In comparison, less than 3% of the untreated cells and cells exposed to the highest dose of H2O2 (1 mM) had taken up PI. Figure 1A shows the appearance of untreated cells under a combination of UV and normal light 10 min after the addition of PI and Hoechst 33342. The nuclear uptake of PI for a proportion of the cells exposed to 100 μM H2O2 for 90 min (Fig. 1B) indicates that cell membrane integrity had been lost. In contrast, control cells (Fig. 1A) and cells exposed to the higher dose of 1 mM H2O2 (Fig. 1C) demonstrate blue nuclear Hoechst staining, indicating that cell membrane integrity was maintained. To provide biochemical confirmation of these findings, the release of lactate dehydrogenase into the culture medium and DNA fragmentation was monitored over a 5-h exposure period to H2O2. Figure 1D shows that at doses of H2O2 over the range 50–200 μM, LDH was released into the culture medium after 90 min exposure to the prooxidant. No significant cellular release of LDH was noted at the higher doses until cells had been exposed to H2O2 for at least 3 h (*, 500 μM, P < 0.0001; **, 1 mM, P < 0.0005; 3 h vs. 30 min). In parallel with the measurement of LDH release, DNA fragmentation was measured with an assay specific for fragmentation that occurs during apoptosis (17). In parallel with the increase in LDH activity at the lower doses of H2O2, Fig. 1E shows that highly fragmented DNA was present after 90 min of treatment. At this time point, no increase in DNA fragmentation was noted at the highest dose of H2O2 (1 mM). After a further 3 h, DNA fragmentation was apparent at the higher doses of H2O2.

    FIG. 1. The effect of exposure of GH3 cells to a dose range of H2O2 over a 5-h time period. The normal light/fluorescence microscopy picture of untreated cells is shown in A, whereas B shows cells treated for 90 min with 100 μM H2O2 and C shows cells treated for 1 mM H2O2 for 90 min. D, The effect of H2O2 exposure on LDH release is demonstrated. E, Over the same concentrations range, apoptotic DNA fragmentation after 90 min and 5 h treatment is shown. For clarity, errors bars are not shown in D, but the maximum coefficient of variation at each data point was not greater than 15% (*, t = 11.21, df = 6, P < 0.0001; **, t = 6.07, df = 6, P < 0.0005; 30 min vs. 3 h). All other data points from 50 μM H2O2 and above at 90 min and 3 and 5 h were significantly different (P < 0.0001) from corresponding 30-min data points. With the exception of 90 min exposure to 1 mM H2O2 (E), all data points were significantly different from control at a level P < 0.001 and above.

    PARP activity and ATP concentrations in response to prooxidant exposure

    To determine the potential role of PARP activation in the response of GH3 cells to H2O2, PARP activity was monitored in response to a dose range of H2O2. After a 45-min exposure to H2O2 (0.05–1 mM), enzyme activity was determined by the addition of [3H]-NAD. Figure 2A shows the incorporation of [3H]-NAD into cells measured as picomoles acid-insoluble material per 106 cells. Only concentrations of 0.5 and 1 mM show incorporation of label significantly greater than that observed for cells untreated with H2O2. When this experiment was repeated over an extended time course of 5 h for exposure to 500 μM H2O2, a significant increase in [3H]-NAD+ incorporation was observed at 1 and 2 h (79 ± 12, t = 16.5, df = 6, P < 0.0001 and 60 ± 18 pmol/106 cells, t = 6.77, df = 6, P < 0.0005 vs. minus H2O2) but not at longer time points. Over a time course of 90 min, Fig. 2B shows the response of ATP to H2O2 at 100 and 200 μM and 1 mM, measured as a percentage of control values. A significant fall in ATP concentrations at 1 mM occurred after 15 min treatment. At lower doses of H2O2, no significant effect on cellular ATP concentrations was noted until cells had been treated for 75 min.

    FIG. 2. The effect of 45 min treatment with a dose range of H2O2 on PARP activity (A) and the response of cellular ATP to a dose range of H2O2 over a time course of 90 min (B). A, Incorporation of [3H]-NAD+ into 1 x 106 cells is shown. Error bars (B) for the 1 mM treatment points at 30 min and more are of the same size as the symbols and therefore have been omitted. Values for 500 μM H2O2 (t = 13.4, df = 6, P < 0.0001) and 1 mM H2O2 (t = 11, df = 6, P < 0.0001) (A) and each indicated (*) data point (B) were significantly different from untreated control samples (t > 5.1, df = 6, P < 0.001 and greater for all samples in B).

    To determine the effect that inhibition of PARP had on cellular ATP concentrations, the experiment presented in Fig. 2B was repeated in the absence and presence of the PARP inhibitors 3-aminobenzamide and nicotinamide (NIC). Figure 3 shows that after 90 min treatment, both PARP inhibitors significantly reduced the fall in ATP concentrations at the high prooxidant doses without affecting values at lower doses.

    FIG. 3. Cellular ATP concentrations in response to a 90-min exposure to H2O2 in the absence and presence of the two PARP inhibitors, 3-aminobenzamide (1 mM) and NIC (1 mM). P < 0.001 vs. H2O2 alone (t > 8.9 with df = 6 for all samples).

    Effect of PARP inhibitors on cell survival

    For this study the effect of NIC alone was investigated. Experiments were performed in which cells were either preexposed to NIC (1 mM) for 1 h before the addition of H2O2 or cells were treated with H2O2 for 1 h and then exposed to NIC (1 mM). For cells preexposed to NIC, Figure 4A shows the proportion of the cell population that had taken up PI after a 2-h exposure to 1 mM H2O2. At doses of 0.5 and 1 mM, a 1-h preexposure to NIC significantly increased the proportion of the cell population that had taken up PI. Indeed, at 0.5 mM this proportion of the cell population is the same as that observed at H2O2 doses around 100 μM. The amount of DNA fragmentation is significantly greater for cells treated with high doses of H2O2 (0.5 and 1 mM) when they had been pretreated with NIC (Fig. 4B).

    FIG. 4. The response of GH3 cells pretreated with 1 mM NIC to a dose range of H2O2 measured as percent PI uptake (*, t = 6.8, df = 6, P < 0.001 vs. H2O2 alone; **, t = 17.4, df = 6, P < 0.0001 vs. control) (A) and as DNA fragmentation (*, 200 mM H2O2: t = 3.7, df = 6, P < 0.005; 500 mM H2O2: t = 16.9, df = 6, P < 0.0001; *, 1 mM H2O2: t = 7.2, df = 6, P < 0.0002 at all indicated points vs. H2O2 alone) (B).

    In contrast to the treatment protocol of Fig. 4, experiments were performed in which after treatment of cells for 1 h with a dose range of H2O2, fresh medium without H2O2 was added. In parallel wells of cells, fresh medium without H2O2 but containing 1 mM NIC was also added to the cells. Metabolic activity was monitored by the addition of the redox sensitive dye, MTT. Microscopic observation of cells at various time points after the addition of H2O2 and in which MTT had been added for 30 min indicated, that within 2 h, cellular metabolism had fallen at around 100–200 μM H2O2; however, at higher doses, metabolic activity was significantly greater. Microscopic observation of these cells (not shown) revealed that all cells with the higher doses of H2O2 maintained a somewhat reduced (compared with untreated cells) ability to metabolize MTT, whereas at the lower doses, a large proportion of the cell population completely failed to metabolize MTT. After a further 2 h, cell swelling was observed at 0.5 and 1 mM H2O2, whereas at the lower doses, there was little change in the proportion of the cell population that failed to metabolize MTT (not shown). Over the next hour, swelling at the high H2O2 concentrations continued, and the addition of MTT (5 h post H2O2) revealed an almost complete loss in ability to metabolize MTT. For cells given NIC, the situation was no different at doses of H2O2 of 200 μM and below; however, at the higher doses, swelling was completely absent and cells maintained an ability to metabolize MTT when added 5 h after the initial administration of H2O2 (see Fig. 5A). Remarkably, for this and other experiments, in the presence of NIC metabolically active (see Fig. 5B), membrane-intact cells excluding PI uptake (not shown) were still apparent even 6 d after first exposure to H2O2. The values for the controls are not given in this figure because cell death had occurred due to medium depletion.

    FIG. 5. The effect of a 1-h pretreatment with a range of H2O2 concentrations followed by prooxidant withdrawal and the addition of fresh medium containing 1 mM NIC. In A, metabolic activity was measured by the addition of MTT 5 h after the first addition of H2O2. In B, metabolic activity was measured 6 d after the first addition of the prooxidant. For all indicated points, t is greater than 5.0, df = 6; *, P < 0.001 and **, P < 0.0001 vs. H2O2 alone.

    The effect of glycolytic and mitochondrial blockade on H2O2-induced death

    Because the delay in cell death at the high H2O2 dose death was paralleled by a large depletion in cellular concentration of ATP, we next tested the effect of blocking glycolytic or mitochondrial ATP synthesis on H2O2-induced cell death. To block mitochondrial function, the protonophore, carbonyl cyanide m-chlorophenylhydrazone (mCCCP; 100 μM) was added to cells 30 min before the addition of a dose range of H2O2. To block glycolysis, medium was changed twice over a 1-h period to DMEM containing 5 mM L-glucose in place of D-glucose before the addition of H2O2. To confirm the loss of mitochondrial function, the cationic dye JC-1 was added to a parallel set of cells 30 min after the addition of mCCCP and also to cells to which media had been changed for L-glucose. Figure 6 shows that changing medium to that containing L-glucose instead of D-glucose had no effect on the proportion of cells with a high m. In contrast, mCCCP resulted in a complete absence of red, J-aggregates and hence a strong reduction of m. Figure 7 shows that the removal of D-glucose markedly enhanced the release of LDH into the culture supernatant 60 min after exposure to the lower doses of H2O2. In contrast, the blockade of mitochondrial function was observed to markedly suppress LDH release at doses of H2O2 over the range 100–250 μM.

    FIG. 6. The response of GH3 cells to the removal of D-glucose (B) or inhibition of mitochondrial function (C) measured by the uptake of the cationic fluorochrome, JC-1, in comparison with control cells (A). Cells with red/orange fluorescence have a high m. Cells with green fluorescence only have a low m.

    FIG. 7. The response of GH3 cell to a dose range of H2O2 in which medium has been replaced with that containing L-glucose or mitochondrial function has been inhibited with the protonophore, mCCCP. LDH release into the culture supernatant was measured after 60 min exposure to a dose range of H2O2 concentrations. All marked (*) values are significantly (t > 10.0, df = 6, P < 0.0001) different from corresponding values at H2O2 alone.

    ROS response of other cells types in presence of NIC

    To compare the response of GH3 cells to other cell types, untransformed HUVECs, transformed vascular endothelial cells (EA.hy 926), and ovarian carcinoma cells (OAW 42) were exposed to a dose range of H2O2 in the absence and presence of NIC. Particularly for the endothelial cells, a morphology change characteristic of necrosis was observed 4 h after exposure to 800 μM H2O2. Limited apoptosis was noted for the HUVECs after exposure to H2O2 (300–400 μM) for 8 h. After 15 h exposure, marked apoptosis was noted for all cell types at doses within the range of 300–600 μM. DNA fragmentation, LDH release, and metabolic activity were measured 18 h after the exposure to H2O2. Figure 8 shows these data. For all cell types, DNA fragmentation increased to reach a maximum over the range 400–500 μM and then fell to control values or below as in the case of HUVECs. This fall corresponded to the light microscope appearance of necrosis, the associated rise in LDH activity detected in the culture supernatant, and the fall in metabolic activity (MTT). A clear finding for all cell types was that NIC added 1 h after H2O2 addition enhanced apoptosis at lower H2O2 doses and markedly blocked necrosis at higher doses; this is shown in Fig. 8 as an increase in DNA fragmentation and a reduction in LDH release, respectively.

    Discussion

    For the majority of cell systems, exposure to high doses of the prooxidant, H2O2, results in rapid (<5 h) necrotic death, whereas at lower doses, apoptosis occurs over a more protracted time frame. In view of this, our studies presented here for a pituitary tumor cell model are unusual because cells die more rapidly at lower doses of H2O2. Although some cell systems undergo rapid apoptotic cell death (4), GH3 cells exposed to lower doses of H2O2 show more of the characteristics of necrosis than of apoptosis. Therefore, cell membrane integrity is lost as evidenced by the uptake of the normally cell impermeable dye, PI, and the release of LDH into the culture supernatant (see Fig. 1). This paradox is further compounded by the finding of DNA fragmentation detected with an assay that has been shown to be specific for DNA fragmentation that occurs during classical apoptosis (17). These and our previous experiments to investigate the mode of death of GH3 cells exposed to H2O2 indicate that the mechanism of cell death is that referred to as necrosis supervened over apoptosis (24). In other words, nuclear apoptosis or a process by which extensive nucleosomal DNA degradation occurs is activated before a catastrophic loss of membrane integrity. Our previous work (17) provided evidence to suggest that in the face of oxidative damage to energy-generating systems, energy-requiring processes are not sufficiently suppressed by the activation of oxidant-sensing mechanisms. This, it has been suggested, leads to an energetic crisis in which ATP levels are insufficient to meet demand and the membrane sodium/potassium pumping mechanism fail. In the current study, we questioned whether this phenomenon is the result of an aberrant stress response and as a strong candidate for a stress-sensing protein, we investigated the role of the enzyme PARP.

    Although the precise cellular role of PARP is not yet fully elucidated, it is generally accepted that the enzyme is involved in the stress response, and in particular it is activated in response to DNA damage (2). ROS are agents that induce DNA damage and are thought to be the principal mediators of radiotherapy-induced DNA damage (7). As a consequence of this, several studies (14, 25, 26, 27) determined the effect of the PARP inhibitor NIC on the response of tumor cells to radiotherapy. The available evidence suggests that NIC markedly enhances radiosensitivity of rodent tumor cells in vivo and that this is not solely due to inhibition of DNA repair. A major substrate for PARP is an endonuclease involved in apoptosis (4). When covalently modified in this way, the endonuclease is inactive; therefore, by inhibiting PARP, NIC will relive inhibition of the endonuclease and allow DNA fragmentation to be initiated. In addition to this, it should be considered that apoptosis is a markedly energetic process requiring ATP (28, 29).

    By inducing PARP activity, NAD+ concentrations are reduced, leading to a fall in ATP concentrations of an extent that blocks apoptosis and in which necrosis becomes the mode of cell death (30). Our observations on the activation of PARP as evidenced by tritiated NAD+ uptake (Fig. 2A) indicate that at the low doses of H2O2, rapid cell death is not due to overactivation of PARP. To the contrary, NAD+ uptake would appear not to be enhanced until cells are exposed to 0.5–1 mM H2O2. This observation is reflected by the changes in cellular ATP concentrations. Within 30 min, 1 mM H2O2 reduces ATP by more than 90%, whereas 100 and 200 μM H2O2 have no effect on ATP concentrations. These observations, coupled with the fact that cell death is delayed at the higher doses of H2O2, suggest that PARP activation offers some degree of protection from oxidant-induced cell death. This idea finds support in the experiments presented in Figs. 4 and 5. By inhibiting PARP with NIC before the addition of H2O2, rapid cell death is induced at the higher doses of 500 μM H2O2. In contrast, when NIC was given after a 1-h exposure to H2O2, a considerable proportion of the cell population remained metabolically viable for a number of days after exposure to highest doses of H2O2. Figure 5B shows this observation for measurements made using the MTT assay. The choice of this assay was due to the significant amount of cell debris, making cell number estimates with a Coulter counter unreliable. Although this assay reflects the cellular nicotinamide adenine dinucleotide phosphate reduced pool, it has previously been shown to correlate with viable GH3 cell numbers (31). In combination, these experiments support the suggestion above that PARP activation in GH3 cells offers some degree of protection against oxidant-induced cell death.

    From the studies presented in this report (see Figs. 2B and 3), it is clear that cell death at the lower doses of H2O2 occurs in the absence of a major fall in cellular ATP concentrations. When the fall in ATP concentration at 0.5 and 1 mM H2O2 is prevented by NIC-induced inhibition of PARP, the effect is to enhance rapid cell death. This suggests that this mode of cell death is responsive to the overall energetic status of the cell. So paradoxically, by enhancing the oxidant-induced fall in ATP concentrations, PARP activation appears to put cells into a state in which cellular processes are all but suspended without a complete inhibition on Na+/K+ATPase activity. Therefore, it would appear that the mechanism by which PARP blocks cell death is by way of its effect on cellular NAD+ levels and in turn, the effect that this has on glycolytic and mitochondrial ATP generation. To provide some confirmatory evidence for this hypothesis, the experiments of Figs. 6 and 7 were performed. Glycolytic activity was inhibited by replacing D-glucose with L-glucose, and mitochondrial activity was inhibited by the protonophore, mCCCP. Measurement of m using the cationic compound JC-1 showed that D-glucose removal had little effect on m, whereas mCCCP completely prevented the appearance of red, J-aggregates, indicating that membrane potential was lost (see Fig. 6). When cells were then exposed to a dose range of H2O2, Fig. 7 shows that mCCCP completely blocked cell death (as measured by the release of LDH activity into the culture supernatant) at the lower prooxidant doses, whereas glucose removal markedly enhanced cell death. These observations unmask a complex dynamic within GH3 cells in which removal of one source of ATP (glycolytic) enhances cell death at the lower oxidant doses and withdrawal of another (oxidative phosphorylation) blocks cell death.

    To more fully integrate these observations into a hypothesis regarding PARP, cell energetics, and oxidant-induced death of GH3 cells, one needs to again consider the hypothesis that was presented at the beginning of this discussion, a hypothesis suggesting that cell death at the lower oxidant doses is due to an inability of cells to mount a stress response by shutting down nonessential energy consuming processes, whereas energy generating processes sustain oxidant damage. So although the steady-state level of cellular ATP may not be greatly altered, a situation might arise in which due to progressive oxidative damage, demand for ATP could momentarily exceed supply, precipitating a catastrophic loss of membrane function and necrosis. As an alternative hypothesis, when metabolic demand is still relatively high, sufficient hexose sugar units may not be available to divert from glycolysis to the pentose phosphate pathway, resulting in a fall in cellular nicotinamide adenine dinucleotide phosphate reduced. This cofactor is essential to maintain the reduced form of glutathione (GSH), the cofactor for the enzyme glutathione peroxidase, the principal agent responsible for reducing H2O2 to water (32). GH3 cells have been previously shown to be extremely sensitive to GSH levels because depletion with the compound L-buthionine sulfoximine has been shown to markedly enhance sensitivity to low doses of H2O2 (17).

    To compare these effects on GH3 cells with other cell types, further experiments were performed on endothelial and cancer cells. Under otherwise similar experimental conditions to those used for GH3 cells, the first signs of cell shrinkage and cellular reorganization (DNA fragmentation as observed by Hoechst 33342 staining; not shown) indicative of apoptosis were observed after around 8 h treatment with 400 μM H2O2. At doses of H2O2 in excess of 600 μM, necrosis was apparent beginning around 4 h after first exposure by the appearance of large membrane swelling and the uptake of PI (not shown). In contrast to GH3 cells, the addition 2 mM NIC, 1 h after initial exposure to H2O2, enhanced the proportion of the cell population undergoing apoptosis at the lower doses and significantly reduced the proportion of the population undergoing necrosis at the higher doses. Indeed, up to 800 μM H2O2, no evidence of necrosis was noted after 4–5 h exposure in the presence of 2 mM NIC. Figure 8 shows a composite of three experiments on the three non-GH3 cell types tested in which LDH activity, DNA fragmentation, and metabolic activity was measure around 18 h after first exposure to H2O2. These data clearly show that NIC enhanced the release of small-sized DNA (apoptotic) nuclear fragments at the lower doses of H2O2 and markedly reduced the release of LDH into the culture supernatant at doses of H2O2 in excess of 500–600 μM. Thus, these cell types demonstrate a normal response to H2O2, in which lower doses induce apoptosis and higher doses induce more rapid necrosis.

    The studies presented above further highlight the unusual nature of the response of these pituitary tumor cells to the prooxidant H2O2. They also provide substantial experimental evidence to implicate the enzyme PARP in this response. For the other cell types described above, high doses of H2O2 (500 μM) induce necrosis and inhibition of PARP by NIC can downgrade the response from necrosis to apoptosis. The difficulty with respect to interpretation of the role of PARP in the death of GH3 cells in response to H2O2 is the unusual nature of cell death induced. Therefore, it has been observed that inhibition of PARP with NIC sensitizes cells to the higher doses of H2O2 by inducing rapid cell death with characteristics of both apoptosis (internucleosomal DNA cleavage) and necrosis (loss of cell membrane integrity). The problem still remains therefore as to why GH3 cells die so rapidly in response to H2O2. Although we have suggested that cells undergo rapid depletion of ATP just before loss of cell membrane integrity, perhaps a more plausible solution is the second hypothesis presented earlier, namely that GSH supply is critically depleted just before loss of cell membrane integrity.

    Whatever the precise mechanism responsible for the death of GH3 cells in response to H2O2, PARP activation would appear to offer a survival advantage. So it would seem that, unlike the other cell systems investigated in this report, GH3 cells are unable to mount a response to lower doses of H2O2 by the activation of PARP. These observations imply therefore that PARP activation serves to reduce the extent of cell death at low doses of ROS. This observation is particularly pronounced for the cancer cell line OAW42, in which NIC markedly enhances sensitivity to H2O2. Given that an endonuclease involved in cleaving DNA during apoptosis is a substrate for PARP (4), our observation for GH3 cells that extensive nucleosomal DNA fragmentation occurs at low prooxidant doses may be due to the inappropriate activation of this endonuclease. Collectively our observations support recent work showing that NIC enhances the sensitivity of glioma cells to a chemotherapeutic agent that damages DNA (33). Furthermore, our results on the treatment of cells with high doses of H2O2 support the idea that PARP inhibitors would be of benefit under conditions such as ischemia-reperfusion in which necrosis is induced by high doses of ROS (34, 35, 36).

    References

    Gaal JC, Pearson CK 1985 Eukaryotic ADP-ribosylation reactions. Biochem J 230:1–18

    Satoh MS, Lindahl T 1992 Role of poly(ADP-ribose) formation in DNA repair. Nature 356:356–358

    Casciola-Rosen L, Nicholson DW, Chong T, Rowan KR, Thornberry NA, Miller DK, Rosen A 1996 Apopain/CPP32 cleaves proteins that are essential for cellular repair: a fundamental principal of apoptotic death. J Exp Med 183:1957–1964

    Schwartzman RA, Cidlowski JA 1993 Apoptosis: The biochemistry and molecular biology of programmed cell death. Endocr Rev 14:133–151

    Richter C 1997 Reactive oxygen and nitrogen species regulate mitochondrial Ca2+ homeostasis and respiration. Biosci Report 17:53–66

    Doroshow JH 1986 Prevention of doxorubicin-induced killing of MCF-7 human breast cancer cells by oxygen radical scavengers and iron chelating agents. Biochem Biophys Res Commun 135:330–335

    Gillies NE 1997 Radiation damage to cell membranes: insights from the oxygen effect. Int J Radiat Biol 71:643–648

    Johnson TM, Yu Z-X, Ferrans VJ, Lowenstein RA, Finkel T 1996 Reactive oxygen species are downstream mediators of p53-dependent apoptosis. Proc Natl Acad Sci USA 93:11848–11852

    Blackstone NW, Green DR 1999 The evolution of a mechanism of cell suicide. Bioessays 21:84–88

    Junod AF, Jornot L, Petersen H 1989 Differential effects of hyperoxia and hydrogen peroxide on DNA damage, polyadenosine diphosphate-ribose polymerase activity and nicotinamide adenine dinucleotide and adenosine triphosphate contents in cultured endothelial cells and fibroblasts. J Cell Physiol 140:177–185

    Watson AJ, Askew JN, Benson RS 1995 Poly(adenosine diphosphate ribose polymerase inhibition prevents necrosis induced by H2O2 but not apoptosis. Gastroenterology 1995:472–482

    Cookson MR, Ince PG, Shaw PJ 1998 Peroxynitrite and hydrogen peroxide induced cell death in the NSC34 neuroblastoma X spinal chord cell line: role of poly (ADP-ribose) polymerase. J Neurochem 70:501–508

    Szabo C, Dawson VL 1998 Role of poly(ADP-ribose) synthase in inflammation and ischemia-reperfusion. Trends Pharmacol Sci 19:287–298

    Horsman MR, Chaplin DJ, Brown JM 1987 Radiosensitization by nicotinamide in vivo: a greater enhancement of tumour damage compared to that of normal tissue. Radiat Res 109:479–489

    Klaidman LK, Mukherjee SK, Hutchin TP, Adams JD 1996 Nicotinamide as a precursor for NAD+ prevents apoptosis in the mouse brain induced by tertiary-butlyhydroperoxide. Neurosci Lett 206:5–8

    Sims JL, Berger SJ, Berger NA 1983 Poly(ADP-ribose) polymerase inhibitors preserve nicotinamide adenine dinucleotide and adenosine 5'-triphosphate pools in DNA-damaged cells: mechanism of stimulation of unscheduled DNA synthesis. Biochemistry 22:5188–5194

    Newton CJ, Drummond N, Burgoyne CH, Speirs V, Stalla GK, Atkin SL 1999 Functional inactivation of the oestrogen receptor by the antioestrogen, ZM 182780, sensitises tumour cells to reactive oxygen species. J Endocrinol 161:199–210

    Newton CJ, Artz E, Stalla GK 1994 Involvement of the estrogen receptor in the growth response of pituitary tumour cells to interleukin-2. Biochem Biophys Res Commun 205:1930–1937

    Jaffe EA, Nachman RL, Becker CG, Minick CR 1973 Culture of human endothelial cells derived from umbilical veins. Identification by morphological and immunological criteria. J Clin Invest 52:2745–2756

    Newton CJ, Eycott K, Green V, Atkin SL 2000 Response of estrogen receptor containing tumour cells to pure antiestrogens and the calmodulin inhibitor, calmidzolium chloride. J Steroid Biochem Mol Biol 73:29–38

    Maellaro E, DelBello B, Comporti M 1996 Protection by ascorbate against apoptosis of thymocytes: implications of ascorbate-induced nonlethal oxidative stress and poly(ADPiribosyl)ation. Exp Cell Res 225:105–113

    Coulter CV, Kelso GF, Lin TK, Smith RAJ, Murphy MP 2000 Mitochondrially targeted antioxidants and thiol reagents. Free Radical Biol Med 15:1547–1554

    Pei-Feng L, Dietz R, von Harsdorf R 1997 Reactive oxygen species induce apoptosis of vascular smooth muscle cell. FEBS Lett 404:249–252

    Raffray M, Cohen GM 1997 Apoptosis and necrosis in toxicology: a continuum or distinct modes of cell death? Pharmacol Ther 75:153–177

    Jonsson G, Kjellen E, Pero RW 1984 Nicotinamide as a radiosensitiser of a C3H mouse mammary adenocarcinoma. Radiother Oncol 1:349–353

    Horsman MR 1995 Nicotinamide and other benzamide analogs as agents for overcoming hypoxic cell radiation resistance in tumours. Acta Oncol 34:571–587

    Rojas A, Denekamp J, John H, Kjellen E, Tsang R, Nilsson P, Stratford MRL, Dennis MF, Joiner MC 1996 Nicotinamide as a repair inhibitor in vivo: studies using single and fractionated X-ray doses in mouse skin and kidneys. Radiation Res 145:419–431

    Eguchi Y, Shimizu S, Tsujimoto T 1997 Intracellular ATP levels determine cell fate by apoptosis or necrosis. Cancer Res 57:1835–1840

    Eguchi Y, Srinivasan A, Tomaselli KJ, Shimizu S, Tsujimoto Y 1999 ATP-dependent steps in apoptotic signal transduction. Cancer Res 59:2174–2181

    Chol Ha H, Snyder SH 1999 Poly(ADP-ribose) polymerase is a mediator of necrotic cell death by ATP depletion. Proc Natl Acad Sci USA 96:13978–13982

    Newton CJ 1995 Estrogen receptor blockade by the pure antiestrogen, ZM 182780, induces death of pituitary tumour cells. J Steroid Biochem Mol Biol 55:327–336

    Brigelius-Flohe R, Banning A, Schnurr K 2003 Selenium-dependent enzymes in endothelial cell function. Antioxid Redox Signal 5:205–215

    Tentori L, Portarena I, Torino F, Scerrati M, Navarra P, Graziani G 2002 Poly(ADP-ribose) polymerase inhibitor increases growth inhibition and reduces G(2)/M cell accumulation induced by temozolomide in malignant glioma cells. Glia 40:44–54

    Kang PM, Haunstetter A, Aoki H, Usheva A, Izumo S 2000 Morphological and molecular characterisation of adult cardiomyocyte apoptosis during hypoxia and reoxygenation. Circulation Res 87:118–125

    Chang ML, Yang J, Kem S, Klaidman L, Sugawara T, Chan PH, Adams Jr JD 2002 Nicotinamide and ketamine reduce infarct volume and DNA fragmentation in rats after ischemia reperfusion. Neurosci Lett 322:137–140

    Khan TA 2003 Poly(ADP-ribose) synthase inhibition improves postischemic myocardial function after cardioplegia-cardiopulmonary bypass. J Am Coll Surg 197:270–277(Christopher J. Newton, De)