Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Although it has been well established that exposure to supraphysiologic concentrations of oxygen inhibits pulmonary cell proliferation, the mechanism leading to this effect remains in question (1). Recent reports suggest that oxidizing agents and oxygen free radicals can indirectly activate cell-cycle regulatory genes by inducing DNA damage, resulting in profound growth inhibition without cell death (4, 5). One of these regulatory genes encodes p21^WAF1/CIP1 protein, a cyclin-dependent kinase (cdk) inhibitor. p21^WAF1/CIP1 is capable of binding to G₁ cyclin-cdk complexes to prevent the phosphorylation of retinoblastoma gene protein and the activation of the E2F pathway necessary for entrance of cells into the S phase of the cell cycle (6). Regulation of the G₁/S transition by p21^WAF1/CIP1 has been linked to both p53-dependent and -independent mechanisms (5). Arrest of the cell cycle at the G₁/S checkpoint is believed to avert replication of damaged DNA templates, which would lead to genomic instability (6, 7). p21^WAF1/CIP1 has also been shown to regulate DNA replication within the S phase in vitro, by binding to proliferating-cell nuclear antigen (PCNA)/cdk complexes and by dissociating cyclin A/cdk2/p107/E2F transitional complexes (8). These interactions block strand elongation by DNA polymerase , and have been shown to occur with and without DNA damage (6, 8, 9).

Recently, it was reported that exposure of mice to 95% O₂ for 72 h induces upregulation of p53 and p21^WAF1/CIP1 messenger RNA (mRNA) and protein in airway epithelium coincidentally with a decrease in cellular bromodeoxyuridine uptake (13, 14). This suggests that oxygen-mediated alterations in gene expression can regulate epithelial cell proliferation, thereby influencing lung growth and development (13, 14). However, whether the changes in gene expression documented in epithelial cells extend to other pulmonary cell types injured by exposure to oxygen, and whether the alterations in regulatory gene expression are associated with specific cell-cycle derrangements, is unknown. Therefore, we conducted the present study to determine the impact of oxygen on actively proliferating human bronchial smooth-muscle cells (BSMC), and to assess whether abnormalities in cell-cycle phase distribution would be associated with increased expression of cell-cycle-regulatory gene products. We hypothesized that oxygen-induced reductions in DNA synthesis in BSMC would be associated with alterations in cell-cycle distribution and with increased expression of p53 and p21^WAF1/CIP1 proteins.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and Culture Conditions

Normal human adult BSMC were obtained from Clonetics (Division of BioWhittaker, Inc., Walkersville, MD) and grown in 6- or 24-well plates and in smooth-muscle growth medium (Clonetics modified Molecular Cellular Developmental Biology [MCDB] medium 131 + 0.5 µg/liter epidermal growth factor + 2 µg/liter fibroblast growth factor-₁ + 5 mg/liter insulin + 50 mg/liter gentamicin + 50 mg/liter amphotericin + 5% fetal bovine serum [FBS]). These cells stain positively for -smooth-muscle actin (Sigma, St. Louis, MO) and display the typical "hill and valley" morphology of smooth-muscle cells when confluent. Passage-5-7 cells seeded at 3,000 cells/cm² were allowed to attach overnight in room air (RA) + 5% CO₂ at 37°C. Human neonatal lung fibroblasts (NLF) were obtained from a bronchial biopsy specimen from an infant with bronchopulmonary dysplasia. NLF were grown as previously described, and stained positively for vimentin but not for smooth-muscle myosin (15). Passage-9-12 NLF grown in M3 medium + 10% FBS were utilized to confirm flow-cytometric findings, and were treated and cultured analogously to BSMC cultures. The effect of oxygen exposure on both BSMC and NLF growth was determined by placing plates into humidified modular incubators (Billups-Rothenberg, Del Mar, CA) flushed with RA, 40% O₂, or 95% O₂ + 5% CO₂ + balanced N₂ at 37°C for 72 h following cell attachment.

Morphology

The effect of oxygen on gross cell morphology was determined with cells grown in four-chamber Labtech slides (Fisher Scientific, Pittsburgh, PA). Cells were fixed in 1% paraformaldehyde and 70% ethanol, and were then stained with hematoxylin and eosin (H&E; Sigma). Morphology was assessed at a magnification of ×200, using an Optiphot microscope (Nikon, Tokyo, Japan).

Proliferation Studies

Monolayers were harvested by incubation with 0.5% trypsin + 1 mM ethylenediamine tetraacetic acid (EDTA) (Sigma) for 2-3 min. Cell number was then determined by counting cells from duplicate chambers with a hemacytometer. DNA synthesis was assessed through thymidine incorporation by cells pulsed for 16 h with 10 µCi/ml [³H]thymidine (Amersham, Arlington Heights, IL). On the morning after the pulse treatment, monolayers were rinsed in phosphate-buffered saline (PBS), and DNA was precipitated by incubating with 10% trichloroacetic acid for 24 h. Samples were then denatured in NaOH and analyzed on a scintillation counter.

Mitochondrial Function

Because alterations in mitochondrial physiology have been linked to both growth arrest and apoptosis, the conversion of 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) to formazen by respiratory enzymes was studied as an index of cellular mitochondrial function (CellTiter 96; Promega, Madison, WI). Conversion was measured by adding 40 µl of MTS per well to samples in 96-well plates, and incubating at 37°C for 4 h. The optical density at 490 nm (OD₄₉₀) was then determined on a Dynatech MR5000 plate reader (Dynatech, Chantilly, VA).

Flow Cytometry

BSMC and NLF monolayers were trypsinized and stained in the unfixed state with the CycleTest Plus DNA kit (Becton-Dickinson, San Jose, CA), using a proprietary combination of trypsin, ribonuclease (RNase) A, and propidum iodide (PI). Suspensions were run on an EPICS Profile II flow cytometer (Phoenix Flow Systems, Hialeah, FL), and the resulting DNA histograms were modeled with Multicycle.AV software (Phoenix).

For the detection of apoptosis, free-floating and attached cells from duplicate T25 flasks were trypsinized and counted. Cells were fixed in 1% paraformaldehyde for 30 min, washed in PBS, and permeabilized with ice-cold 70% ethanol. After rinsing, 3' ends of DNA cleavage sites were labeled with the terminal deoxynucleotidyl transferase-dUTP nick-end labeling (TUNEL) technique, using a commercially available kit (APO-Direct, Phoenix) (16). Cells were incubated for 60 min at 37°C in terminal deoxyuridine transferase/fluorescein-dUTP, rinsed, and incubated with PI/RNase A for 30 min. Flow cytometry was conducted with gating parameters established on campothecin-treated HL-60 cells supplied by Phoenix.

Lactate Dehydrogenase Release

Cell death was determined through released lactate dehydrogenase (LDH) (1). Media-derived and cellular LDH were measured colorimetrically, using pyruvate as a substrate and NAD/oxidase as a detector (Sigma). High- and low-concentration LDH standards (Sigma) were run during each assay to validate the LDH calibration curve. Cellular LDH content was determined on scraped cells that had been sonicated for 2 min. The percent LDH release was then determined by dividing the content of LDH in the medium by the total LDH per well.

Enzyme-Linked Immunosorbent Assay for p53 and p21^WAF1/CIP1

Free-floating and attached cells were counted on a hemacytometer prior to assaying for p53 and p21^WAF1/CIP1 with enzyme-linked immunosorbent assays (ELISA) (Oncogene, Cambridge, MA). Briefly, cells were centrifuged and resuspended in 200 µl of 10 mM Tris + 5 mM EDTA + 0.2 mM phenylmethylsulfonyl fluoride + 1 µg/ml pepstatin + 0.5 µg/ml leupeptin (pH, 7.4). Antigen was extracted by adding 40 µl of extraction reagent (Oncogene) to the cell suspension and incubating on ice for 30 min. Determination of peptide concentrations was done by adding 100 µl of cell lysate to 96-well plates preabsorbed with either mouse monoclonal anti-p53 or rabbit polyclonal anti-p21^WAF1/CIP1 antibody. Plates were incubated for 2 h, washed, and incubated for 1 h with 100 µl biotinylated mouse monoclonal anti-p53 or p21^WAF1/CIP1 antibody. The presence of secondary antibody was detected colorimetrically through the use of streptavidin-horseradish peroxidase, with tetramethylbenzidine as the chromogen. Color change was read at 450 /540 nm on a Dynatech plate reader (Chantilly, VA), and corresponding peptide concentrations were extrapolated from a standard curve produced from known concentrations of p53 and p21^WAF1/CIP1 peptide supplied by the ELISA manufacturer. The sensitivities of the assays as reported by the manufacturer are 0.012 ng/ml for p53 and 0.1 U/ml for p21^WAF1/CIP1 (1 U = [p21^WAF1/CIP1 from 2.7 × 10⁵ MCF7 cells]). Values derived from the standard curves were normalized to 10⁴ cells.

Statistics

All parameters, except apoptosis, were analyzed through repeated measures analysis of variance (ANOVA), followed by pairwise comparisons made with Newman-Keuls test, using GB-STAT software (Dynamic Microsystems, Silver Springs, MD). Apoptosis data were rendered descriptively. Reported values represent mean ± SD, with the level of significance set at P  0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Effect of Oxygen on Cell Morphology

BSMC exposed to 95% O₂ displayed progressive long-axis shortening over time, with a generalized loss of the spindle-shaped phenotype. Cells exposed to 40% O₂ appeared similar in phenotype to control cells, but were fewer in number at 48 and 72 h (Figure 1).

View larger version (128K): [in this window] [in a new window]   Figure 1.   Effect of oxygen on BSMC morphology. Photomicrographs of BSMC grown on Labtech slides and exposed to 72 h of RA (a), 40% O2 (b), or 95% O2 (c). Cells were fixed in paraformaldehyde and ethanol, and were stained with H&E. Note progressive decrease in cell number and spindle-axis shortening with increasing oxygen concentration.

Effects of Oxygen on Cell Proliferation and Mitochondrial Function

Exposure of BSMC to 95% O₂ led to a rapid decrease in cell proliferation, with no significant change in cell number occurring beyond 24 h (Figure 2a). Although cells exposed to 40% O₂ also showed decreased proliferation as compared with RA-treated cells, the net effect was not apparent until 72 h. As would be expected from the cell-number data, DNA synthesis was negatively affected by increasing oxygen concentration and exposure (P < 0.0001) (Figure 2b). At 72 h, DNA synthesis in 40% O₂- and 95% O₂-treated BSMC was reduced to just 22% and 7% of RA cells, respectively.

View larger version (24K): [in this window] [in a new window]   Figure 2.   Effect of oxygen on BSMC proliferation and DNA synthesis. (a) Effect of oxygen on proliferation as assessed through direct cell counts. Cell number was negatively effected by increasing oxygen concentration and duration of exposure (ANOVA, P < 0.0001). Counts from plates exposed to 40% O2 were lower than those from RA-exposed cells at 72 h (+P < 0.01). Cells exposed to 95% O2 were fewer in number than those exposed to RA or to 40% O2 at 24-72 h (*P < 0.01). Data represent means of six replicates. (b) Effect of oxygen on DNA synthesis as assessed through [3H]thymidine incorporation. DNA synthesis was also negatively affected by increasing oxygen concentration and duration of exposure (ANOVA, P < 0.0001). Thymidine incorporation was greater from 24-72 h in RA-exposed cells than in either 40% O2- or 95% O2-exposed cells (+P < 0.01). Cells treated with 95% O2 also incorporated less thymidine than cells treated with 40% O2 beyond time-0 (*P < 0.01) (dpm = disintigrations/minute). Data represent mean of 12 replicates, with bars indicating SD.

In addition to its effects on cell division, oxygen altered mitochondrial function in a concentration- and time-dependent manner (P < 0.0001) (Figure 3). The results reveal that 95% O₂ reduced the conversion of MTS to formazen within 24 h, whereas 40% O₂ appeared to have little effect until 48 h. Furthermore, MTS conversion in 95% O₂-treated cells was below the baseline value at each of the subsequent 24-h periods (time-0 versus 24, 48, and 72 h; P < 0.05).

View larger version (23K): [in this window] [in a new window]   Figure 3.   Effect of oxygen on BSMC mitochondrial function. Mitochondrial function of BSMC was determined from the conversion of MTS to formazen, as detected colorimetrically at 490 nm (OD490). Mitochondrial function was negatively affected by increasing oxygen concentration and duration of exposure (ANOVA, P < 0.0001). OD490 values for cells exposed to 95% O2 were less than that of RA- and 40% O2-exposed cells at 24-72 h (*P < 0.01). Cells treated with 40% O2 converted less MTS to formazen than did RA-exposed cells at both 48 and 72 h (+P < 0.01). Data represent means of 12 replicates, with bars indicating SD.

Effect of Oxygen on Cell-Cycle Distribution and Cell Death

In BSMC, all phases of the cell cycle were altered by oxygen exposure, although the distribution pattern of 95% O₂-treated cells changed little over the 72-h period (Figures 4 and 5). As compared with exposure to RA, exposure to 95% O₂ caused a marked accumulation of cells in the S-phase and a reduction of cells in the G₂/M phase at 24 and 48 h. By 72 h, not only were a greater percentage of 95% O₂-treated cells in S phase (RA: 12.9 ± 4.6%, versus 95% O₂: 34.6 ± 7.0%, P < 0.01), but also a smaller fraction of 95% O₂-treated cells than cells exposed to RA were in the G₀/G₁ phase (RA: 76.4 ± 8.7%, versus 95% O₂: 59.3 ± 6.0%; P < 0.01). By contrast, exposure to 40% O₂ changed the cell-cycle distribution little until 72 h, at which time there appeared to be a shift in distribution into the G₂/M phase. To assess whether the alterations in cell-cycle distribution were unique to BSMC, we exposed NLF to room air and 95% O₂ under identical conditions. Once again, we found that exposure to 95% O₂ produced an accumulation of cells in the S phase on each of the three days of exposure, which was more than 2-fold greater than RA-treated cells at 72 h (RA: 12.6 ± 1.8%, versus 95% O₂: 28.8 ± 1.9%; P < 0.01). As with BSMC, S-phase changes were accompanied by reductions in the fraction of cells in the G₀/G₁ and G₂/M phases.

View larger version (18K): [in this window] [in a new window]   Figure 4.   Representative DNA histograms obtained after 72 h of exposure of BSMC to RA, 40% O2, and 95% O2. Histograms were produced by flow-cytometric analysis of nonfixed, propidium iodide-stained cells. Histograms were modeled with Multicycle.AV software and the aggregates paradigm. Note the large accumulation of 95% O2-exposed cells in the S phase.

View larger version (31K): [in this window] [in a new window]   Figure 5.   Effect of oxygen on BSMC cell-cycle distribution, as determined with propidium iodide. Exposure to 95% O2 for 72 h reduced the percentage of cells in G0/G1 as compared with RA-treated cells (**95% O2 versus RA, P < 0.01). In addition, the fraction of 95% O2-exposed cells in S phase was greater at 24- 72 h than that of RA- or 40% O2-exposed cells (*95% O2 versus RA and 40% O2, P < 0.01). In G2/M, there was a reduction in 95% O2-exposed cells at 24 and 48 h (*95% O2 versus RA and 40% O2, P < 0.01) and an accumulation of 40% O2-exposed cells at 72 h (+40% O2 versus RA and 95% O2, P < 0.01). Columns represent means of seven replicates and bars indicate SD.

We found a small but significant increase in percent LDH release from all cell groups over time (range: 2.1% at 24 h; 4.4% at 72 h; P < 0.05). Nevertheless, there were no differences in the percent LDH released among cells exposed to the three oxygen concentrations at any time point. Using TUNEL labeling, we were able to detect apoptosis in 23.8% of the camptothecin-treated HL-60 cells. In BSMC, however, the percentage of apoptotic cells was extremely low at all times and under all conditions, being approximately 0.3% at time-0 and less than 0.1% thereafter.

Effect of Oxygen on p53 and p21^WAF1/CIP1

p53 and p21^WAF1/CIP1 proteins were detected in all cell groups at all time points, and each value exceeded the known lower limit of detection of the ELISAs for these proteins by at least a factor of five. Both p53 and p21^WAF1/CIP1 were affected by oxygen and by the duration of exposure (P < 0.05). As shown in Figure 6a, p53 levels tended to decrease over time. Nevertheless, p53 protein per cell was greater in 95% O₂-exposed than in RA-exposed cells throughout the study period, with significant differences being observed at 24 and 72 h. The levels of p21^WAF1/CIP1, on the other hand, remained constant during the 72-h study in both RA- and 40% O₂-exposed BSMC (Figure 6b). Cells exposed to 95% O₂, however, showed a generalized increase in p21^WAF1/CIP1 protein/cell over time, which was maximal at 48 h (RA: 0.57 ± 0.20, versus 95% O₂: 1.29 ± 0.26 U/ml/10⁴ cells; P < 0.01).

View larger version (30K): [in this window] [in a new window]   Figure 6.   Effect of oxygen on p53 and p21WAF1/CIP1 protein levels in BSMC. (a) Effect of 72 h of oxygen exposure on p53 levels as determined through ELISA. Levels of p53 increased with increasing oxygen concentration but decreased overall with time in all atmospheres (ANOVA, P < 0.05). Expression of p53 was greater in 95% O2-exposed cells than in 40% O2- and RA- exposed cells at 24 and 72 h (*P < 0.01). (b) Effect of 72 h of oxygen exposure on p21WAF1/CIP1 levels as determined through ELISA. Levels of p21WAF1/CIP1 increased with increasing oxygen concentration (ANOVA, P < 0.0005), with individual differences noted at 48 h (*95% O2 versus 40% O2 and RA, P < 0.01). Columns represent means of seven replicates and bars indicate SD.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Numerous studies have demonstrated the ability of oxygen to slow the proliferation of cells grown in culture. Alveolar type II cells, fibroblasts, and tracheal smooth-muscle cells all exhibit decreased growth within 24 h of exposure to O₂ fractions exceeding 50% O₂ (1, 16). This high-oxygen environment increases the generation of reactive oxygen species (ROS) and fosters the release of H₂O₂ from mitochondria, either of which effect may lead to DNA strand breaks, protein oxidation, and lipid peroxidation (17, 18). Although an abundance of literature has focused on the nucleus as a central target of oxidative damage, it is clearly not the only organelle injured by oxygen. Yakes and Van Houten recently demonstrated that ROS-induced injury of mitochondrial DNA is more extensive, and repair less rapid, than for nuclear DNA (19). In addition, hyperoxia has been shown to diminish mitochondrial energy production by inactivating key respiratory enzymes such as aconitase and reduced nicotinamide adenine dinucleotide-oxidase (20, 21). These alterations may then promote a stress response that culminates in either growth arrest or apoptosis. Because of the intimate involvement of mitochondria in proliferative and apoptotic signaling, we examined the impact of hyperoxia on mitochondrial function, using the bioreduction of MTS. We found that exposure of BSMC to 95% O₂ rapidly and profoundly reduced DNA synthesis without inducing apoptosis, a finding that has previously been made in hyperoxia-exposed A549 cells (22). At the same time, we also found a marked depression in mitochondrial respiration in 95% O₂. By exposing cells to a lower oxidative load, however, we were able to show that reductions in DNA synthesis precede reductions in MTS absorbance, indicating that oxygen-mediated inhibition of mitochondrial energy production is unlikely to instigate the loss of DNA synthetic capacity.

The cessation of proliferation observed during oxidative stress is believed to represent a basic protective mechanism that prevents the replication of damaged template DNA. Classically, cell-cycle arrest occurs at one of two well-defined checkpoints: the G₁/S transition or the G₂/M transition (23). Nevertheless, we found that both BSMC and NLF grown in 95% O₂ accumulated to the greatest degree within the S phase, a response that appeared early and was sustained throughout the study period. Although unexpected, this finding is consistent with the existence of a damage-sensitive checkpoint within the S phase. In yeast, this regulatory site involves a complex signal-transduction pathway initiated by the recognition of damage by DNA polymerase  and propagated through the activation of the lipid and protein kinase genes MEC1 and RAD53 (24, 25). An analogous pathway in mammalian cells involves the ataxia-telangiectasia-mutated (ATM) gene (26). The ATM gene has been found to have oxidative stress- and DNA damage-sensing properties, and to regulate cell-cycle progression (26, 27). The gene not only regulates the G₁/S and G₂/M phase transitions by activating p53 and p21^WAF1/CIP1, but also controls S-phase traversal by altering the DNA-binding capacity of replication protein A (26, 27). In essence, the ATM gene serves a managerial function for the cell, coordinating injury detection, cell-cycle progression, and repair.

Throughout the cell cycle, progression is controlled by a series of cyclins and cdks that are in turn regulated by cdk inhibitors such as p21^WAF1/CIP1, p27^kip1, p16, and others (6, 28, 29). In normal cells, p21^WAF1/CIP1 exists in quaternary complexes with cyclins, cdks, and PCNA (6, 30). Overexpression of p21^WAF1/CIP1 alters the stoichiometric relationship within the complex, leading to growth arrest either through the inactivation of E2F transcriptional factors in late G₁ or through the dissociation of cyclin A from the p107/cdk2/E2F transitional complex in the S phase (8, 10, 12). Although the observation that hyperoxia-treated BSMC accumulate in the S phase and show increased p21^WAF1/CIP1 expression implies participation of p21^WAF1/CIP1 in S-phase regulation, the association may not be causal. A recent study of diploid fibroblasts found that bleomycin-induced DNA damage increased p53 and p21^WAF1/CIP1 protein expression coincidentally with G₁/S- and S-phase arrest. Transfection of these cells with a dominant-negative mutant of p53 abolished the p21^WAF1/CIP1 response and the G₁/ S blockade, but did not augment DNA synthesis, suggesting that damage sufficient to impair DNA synthesis occurred independently of the p53 surveillance mechanisms regulating p21^WAF1/CIP1 expression (31). Furthermore, the induction of DNA strand breaks in colon carcinoma cells, and the subsequent S-phase arrest of these cells, has been shown to precede the upregulation of p53 and p21^WAF1/CIP1 proteins, indicating that the arrest is secondary neither to p53-mediated mechanisms nor to p21^WAF1/CIP1-induced inactivation of DNA-replicating machinery (32). Additionally, it is conceivable that p21^WAF1/CIP1 serves to protect cells from p53-mediated apoptosis, a notion supported by the report that ectopic expression of p21^WAF1/CIP1 protects p21^WAF1/CIP1-deficient mouse embryonal fibroblasts from p53-mediated apoptotic pathways (33). Thus, during extreme oxidative stress, the upregulation of p21^WAF1/CIP1 may initiate growth arrest not only to allow time for cellular repair of damaged DNA, but also to avert premature cell death.

Although our observations contribute to the understanding of oxygen toxicity and cell-cycle regulation, their direct application to oxygen-related disease is less clear. The typical pathophysiologic response of airway smooth-muscle cells to hyperoxia in vivo is proliferative, driven by the stimulation of mitogens released from inflammatory cells, epithelial cells, and fibroblasts (34). Because the levels of oxidative stress utilized in the present study exceeded those experienced by cells in the airway wall, one could argue that oxygen-induced growth arrest without apoptosis is purely an in vitro phenomenon. This conclusion is strengthened by the knowledge that hyperoxia induces apoptosis in the airway epithelium of mice but necrosis in cultured A549 cells, a dichotomy that most likely reflects the influence of cell-cell interactions and inflammatory mediators in vivo (22, 35, 36). Nevertheless, studies in the rat indicate that the replicative response of airway smooth-muscle cells to 95% O₂ is dynamic and follows a pattern similar to that of the airway epithelium (37, 38). Although data are lacking on the acute smooth-muscle response to hyperoxia, it would seem reasonable to postulate that airway smooth-muscle cells, like their epithelial counterparts, experience a reduction in proliferation during this period (14). Subsequent mitogenic stimulation might then induce a relative overexpression of cyclins A and D₁, prompting the return of cell-cycle progression, a process described in fibroblasts constitutively expressing p21^WAF1/CIP1 and in alveolar epithelial cells recovering from hyperoxic injury (10, 39).

In conclusion, we have shown that oxygen produces graded reductions in DNA synthesis and mitochondrial function in BSMC, and that extremely high concentrations of oxygen induce an accumulation of cells in the S phase of the cell cycle. These changes occur in conjunction with increased expression of p53 and p21^WAF1/CIP1 proteins, but without significant cell loss, suggesting that upregulation of cell-cycle regulatory genes occurs to promote cellular repair rather than apoptosis.