Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The mammalian lung is chronically exposed to oxidant gases because it functions to exchange oxygen and carbon dioxide between the environment and blood. It is therefore at risk from oxidant-induced injury caused by hyperoxia or hyperbaric oxygen used to treat tissue hypoxia, as well as by inhaled particles, ozone, and other environmental pollutants. Although efforts are made to reduce pollutant exposure in the environment, supplemental oxygen is a common clinical intervention for newborns, children, and adults with respiratory distress. Unfortunately, oxygen levels > 90% cause extensive swelling and necrosis of alveolar endothelial and type I epithelial cells, resulting in the formation of hyaline membranes and increased morbidity and mortality (1). In contrast, the type II epithelial cell is relatively resistant to hyperoxia and its survival is essential for normal repair during recovery in room air because it repopulates dead type I cells through proliferation and differentiation (2). Although it remains unclear how hyperoxia injures and kills cells, it is believed to be converted to cytotoxic reactive oxygen species (ROS) of hydrogen peroxide, superoxide anion, and hydroxyl radicals that damage DNA, protein, and lipids.

Recent studies suggest that the response to DNA damage may be important because DNA fragmentation and increased expression of the tumor suppressor protein p53 and its downstream target genes have been observed in murine lungs exposed to hyperoxia (3). p53 increases in cells with damaged DNA and induces genes involved in growth control, DNA repair, and apoptosis (8). A major target of p53 is the cyclin-dependent kinase inhibitor p21^{Cip1/WAF1/Sdi1} (p21), which inhibits proliferation and promotes DNA repair. Alternatively, p53 can induce cell death by increasing expression of the proapoptotic Bcl-2 family member Bax. Although mouse lungs exposed to severe hyperoxia have increased expression of p53 (3, 4), p21 (5), and Bax (3), their role in lung injury has yet to be fully clarified. Several investigators have exposed p53- deficient mice to hyperoxia in an effort to understand its role in the injury process. One study found that p21 messenger RNA (mRNA) was induced in a p53-dependent manner when adult mice were exposed to 92 to 93% FiO₂ (6). Unfortunately, the effect of hyperoxia on proliferation, cell injury and survival was not assessed in these mice. Other studies demonstrated that p53 deficiency does not modify hyperoxic lung injury as assessed by terminal deoxyribonucleotidyl transferase-mediated deoxyuridine triphosphate-biotin nick-end labeling (TUNEL) staining or wet-to-dry (W/D) lung ratios, an indicator of edema (3, 9). These studies did not determine whether p53 deficiency altered expression of p21 or Bax. Thus, conclusions that p53 deficiency does not modify lung injury cannot be extrapolated to include its target genes without detailed analysis of their expression.

p21 is an attractive candidate to modify lung injury because in vitro studies have revealed that it protects cells against death caused by exposure to ultraviolet (UV) radiation and alkylating agents such as cisplatin or nitrogen mustard (10). p21 was independently identified by several groups as a p53-regulated protein that inhibits cell growth when transfected into cells (WAF1) (11), as a cyclin-dependent kinase (Cdk) 2-interacting protein (Cip1) (12), and as a gene whose expression increases in senescent cells (Sdi1) (13). Subsequent studies revealed that p21 is also regulated independent of p53 by the cytokine transforming growth factor (TGF)-, interleukin (IL)-6, and differentiation (14- 16). In addition, oxidative stress, induced by compounds such as diethylmaleate, can increase p21 expression through a p53-independent mechanism (17). The p21 protein may be functionally separated into an amino-terminal fragment that inhibits G1- and S-phase Cdk activities and a carboxy-terminal fragment that binds proliferating cell nuclear antigen (PCNA) (18). Although both interactions inhibit DNA replication, the p21-PCNA interaction does not inhibit PCNA-dependent DNA repair in vitro (19). Collectively, these studies reveal that p21 is an important regulator of the cellular response to genotoxic stress.

p21 is likely to mediate the growth-arresting activities of hyperoxia because its expression increases when proliferation decreases (5, 20). Moreover, p21 may enhance survival by promoting DNA repair or modifying cell death and survival. In the present study, we use p53- and p21-deficient mice to show that 100% oxygen induces p53-independent expression of p21 that markedly enhances survival and inhibits DNA replication. These findings reveal an important role of p21 to protect pulmonary cells from oxidative stress caused by in vivo exposure to hyperoxia.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Mice and Exposure Conditions

Normal adult (8 to 12 wk) pathogen-free male C57Bl/6J mice were obtained from Jackson Laboratories (Bar Harbor, ME). p53 (+/+) and p53 (/) C57Bl/129 hybrid mice were obtained from Taconic (Germantown, NY). p21 (/) mice were obtained from Dr. Anita Roberts (NCI, NIH, Bethesda, MD) with the permission of Dr. Philip Leder (Harvard Medical School, Cambridge, MA) (21). The response to hyperoxia was assessed with C57Bl/6J mice because the p21-null allele was backcrossed several generations onto this background, which is considered to be a sensitive strain (22). Mice were kept in room air or exposed to > 95% oxygen by placing the cages inside a Plexiglas chamber through which prewarmed, humidified, and filtered oxygen was delivered via a 0.22-µm filter (4, 5). Mice were injected with 5-bromo-2'-deoxyuridine (BrdU) as recommended by the manufacturer (Zymed, South San Francisco, CA) 2 h before the animals were killed with pentobarbital (65 mg/kg, injected intraperitoneally). The lungs were exposed and the right lobes ligated and removed for isolation of RNA or protein. The left lobe was inflation-fixed through the trachea and sectioned for TUNEL staining, in situ hybridization, or ultrastructural analysis. Some mice were lavaged 10 times with saline and total protein was quantified using a modified Lowry assay (Bio-Rad, Hercules, CA) with bovine serum albumin as a standard. The total number of cells in the bronchoalveolar lavage fluid (BALF) was quantified in the first lavage using a hemacytometer and normalized to volume. W/D lung ratios were determined by weighing the left lobe before and after desiccation at 80°C and normalizing to total body weight. The University of Rochester's University Committee on Animal Resources approved all exposures and handling of the mice.

Northern Analysis

Total RNA was separated on 1.0% agarose-formaldehyde gels and transferred to Nytran. Blots were hybridized with a ³²P-labeled 454-base pair (bp) complementary DNA (cDNA) containing the second exon of the mouse p21 gene (21). A 553-bp cDNA for histone H3 was amplified from mouse lung RNA by reverse transcriptase/polymerase chain reaction (RT-PCR) using sense primer 5'-AGCAGAGGCTGACCAATCCCAACAAAGCG and antisense primer 5'-TCGTTTAAGCCCTCTCCCCACGAATGCG. Due to the high GC content of histone H3, PCR was performed using the Advantage-GC 2 PCR Kit (Clontech; Palo Alto, CA). The amplified cDNA was purified, blunt-ended, and inserted into the pCR ScriptAmp SK+ cloning vector using the Stratagene PCR ScriptAmp Cloning Kit (Stratagene; La Jolla, CA). Blots were hybridized with ³²P-labeled antisense H3 RNA generated by in vitro transcription. Changes in gene expression were normalized to expression of mRNA for the ribosomal subunit L32 or 18S RNA by PhosphorImage analysis (5).

Immunoblotting

Lungs were homogenized in Tris-buffered saline containing 0.2% triton X-100, 0.3% Nonidet P-40, and protease inhibitors (4). Proteins (50 µg/ml) were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis, transferred to Nitrocellulose membranes, and blotted with polyclonal CM5 (Novacastra, Newcastle, UK), directed against p53, or anti--actin (Sigma, St. Louis, MO). Bound antibody was detected with chemiluminescence (Amersham, Arlington Heights, IL).

In Situ Hybridization

Antisense and sense ³³P-radiolabeled riboprobes were synthesized using T3 and T7 RNA polymerases to a specific activity of 3 × 10⁹ disintegrations per min/µg. In situ hybridizations were performed on 4-µm sections of mouse lung as described previously, with the following modification (5). Sections were prehybridized for 3 h and hybridized for 16 h at 53°C. After washes and digestion with RNAse A, sections were washed stringently in 0.1× saline sodium citrate for 30 min at 68°C, dipped in a 1:1 dilution of NTB-2 emulsion (Eastman Kodak, Rochester, NY) and exposed at 4°C before developing and counterstaining with hematoxylin and eosin (H&E). Slides were visualized with a Nikon E800 microscope (Nikon, Melville, NY) and images captured with a SPOT-RT digital camera (Diagnostic Instruments, Sterling Heights, MI).

DNA Fragmentation

TUNEL staining was performed using an ApopTAG kit from Oncor (Gaithersburg, MD). TUNEL-positive cells stained brown due to reaction with 3,3'-diaminobenzidine (Sigma). As we reported previously, TUNEL staining was not detected in sections reacted in the absence of terminal transferase (4).

Statistical Analysis

Values are expressed as means +/ standard deviation. Group means were compared by analysis of variance with Fisher's procedure post hoc analysis using StatView software for Macintosh, with P < 0.05 considered significant. Kaplan-Meier analysis was used to estimate survival function.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Hyperoxia Increases p53 and p21 Expression

Previous studies demonstrated that hyperoxia increased p53 and p21 expression in terminal bronchioles and throughout the parenchyma of the adult mouse lung (4, 5). p53 (+/+) and p53 (/) mice were exposed to room air or hyperoxia to determine whether induction of p21 was dependent upon p53. p21 mRNA was detected faintly in p53 (+/+) mice exposed to room air and significantly increased relative to L32 after exposure to hyperoxia (Figure 1). p21 mRNA was also detected faintly in p53 (/) mice exposed to room air and significantly increased in hyperoxia, but to a lesser extent than observed for p53 (+/+) mice. After 84 h of exposure, p21 mRNA had increased in p53 (/) lungs to a level similar to that detected in p53 (+/+) lungs (data not shown). We confirmed that the samples were derived from p53 (/) mice and not p53 (+/) by RT-PCR as described by Donehower and colleagues (data not shown) (23). In situ hybridization revealed that hyperoxia increased p21 in terminal bronchioles and throughout the parenchyma in p53 (+/+) lungs, as previously reported (5), and in p53 (/) lungs (data not shown). Thus, in vivo exposure of lungs to hyperoxia increases p53-independent expression of p21.

View larger version (26K): [in this window] [in a new window]   Figure 1.   Hyperoxia increases p21 in the absence of p53. (A) Total lung RNA from p53 (+/ +) and p53 (/) mice exposed to room air (0) or > 95% oxygen for 48 h was hybridized for p21 and L32. (B) The intensity of hybridization was quantified by PhosphorImage analysis and graphed relative to values obtained from p53 (+/+) mice exposed to room air. Hyperoxia significantly increased p21 mRNA in p53 (+/+) mice (*P < 0.0001) and p53 (/) mice (*P < 0.05). p21 induction was less in the p53 (/) mice compared with p53 (+/+) mice. Numbers of mice used for analysis are above each bar.

Previous studies demonstrated that p53 protein increased 3-to 5-fold in lungs of adult mice exposed to hyperoxia (3, 4). Western blot analysis was used to assess whether hyperoxia induced p53 in the absence of p21. As expected, p53 was detected in homogenates prepared from p21 (/) lungs exposed to room air and increased approximately 5-fold after exposure to 48 h of hyperoxia (Figure 2). Hyperoxia therefore increases p53 in the absence of p21.

View larger version (23K): [in this window] [in a new window]   Figure 2.   Hyperoxia increases p53 in the absence of p21. (A) p21 (/) mice were exposed to room air (0) or hyperoxia for 48 h and lung homogenates were blotted for p53 and -actin. (B) p53 expression was normalized to -actin levels and graphed relative to values obtained from room air- exposed lungs. Hyperoxia increased p53 abundance (*P = 0.05). Numbers of mice used for analysis are above each bar.

p21 Deficiency Reduces Survival Time and Increases Edema

Although other studies found that p53 deficiency did not modify hyperoxic lung injury (3, 9), our finding of increased p21 during hyperoxia independent of p53 suggests that p21 may modify the lung response. We therefore decided to assess lung injury and survival in p21-deficient mice. p21 (+/+) mice died between 103 and 126 h of exposure, with a mean survival time of 120 +/ 2.25 h (Figure 3A). In contrast, p21 (/) mice rapidly succumbed to hyperoxia between 60 and 81 h, with a mean survival time of 72 +/ 2.2 h. W/D lung ratio and total amount of protein in BALF was used to assess the degree of edema. The W/D ratios of p21 (+/+) and (/) lungs exposed to room air were not significantly different (Figure 3B). This ratio did not increase significantly in p21 (+/+) mice exposed to hyperoxia for 72 h. In contrast, the W/D ratio significantly increased 50% in p21 (/) lungs exposed to hyperoxia. Similarly, BALF protein was not significantly different between p21 (+/+) and (/) mice exposed to room air (Figure 3C). In contrast, hyperoxia increased BALF protein levels 3.5-fold in p21 (+/+) lungs, whereas it caused a significantly greater 9.8-fold increase in p21 (/) lungs. Thus, there was significantly more edema in p21 (/) lungs compared with p21 (+/+) lungs after similar times of exposure to hyperoxia, consistent with the earlier onset of mortality.

View larger version (27K): [in this window] [in a new window]   Figure 3.   Survival and edema of mice exposed to hyperoxia. (A) Survival curve of mice exposed continuously to hyperoxia. Statistical analysis was performed using Wilcoxon two-sample test assuming nonparametric data (n = 12 for each genotype, P < 0.0001). (B) Hyperoxia increased the W/D lung ratios of p21 (/) lungs (n = 3 for each genotype, *P < 0.01). (C) Hyperoxia increased BALF protein p21 (+/+) lungs (*P < 0.01) and p21 (/) lungs (*P < 0.005). BALF protein was also significantly greater in hyperoxic p21 (/) lungs compared with hyperoxic p21 (+/+) lungs. (D) Cell number in BALF was not altered by hyperoxia in p21 (+/+) mice, but was significantly decreased in p21 (/) mice (*P < 0.002). Values represent number of cells per milliliter of lavage fluid. Numbers of mice used for analysis are above each bar (B-D).

Because hyperoxia induces an inflammatory response that could promote cell injury and death, we assessed whether sensitivity of the p21 (/) mice was due to an overly robust inflammatory response. Inflammatory cells were quantified in BALF obtained from mice exposed to room air and 72 h of hyperoxia. The number of cells obtained from p21 (+/+) mice exposed to room air was not significantly different from comparably exposed p21 (/) mice (Figure 3D). Although lavagable cell number did not increase in p21 (+/+) lungs exposed to hyperoxia, consistent with the fact that they were not severely injured yet, it was significantly decreased in p21 (/) lungs. This suggested that the mice failed to mobilize an inflammatory response or that it was difficult to isolate cells from the lungs. The latter hypothesis is more likely because abundant inflammatory cells were not observed upon histologic examination (see Figures 4 and 5). The expression of the proinflammatory cytokines IL-1 and IL-6 were also measured by ribonuclease protection analysis because hyperoxia induces their expression (22). Although their expression increased modestly after 72 h of hyperoxia, there was no significant difference between p21 (+/+) and p21 (/) mice (data not shown). Further, we challenged the mice with lipopolysaccharide to induce an inflammatory response, and measured mRNA levels of several chemokines, including eotaxin, monocyte chemotactic protein-1, macrophage inflammatory protein (MIP)-1 and MIP-1. Again, we did not detect any difference in the inflammatory response between p21 (+/+) and p21 (/) mice (data not shown). Our findings are consistent with the hypothesis that the sensitivity of the p21 (/) mice to hyperoxia is not due to alterations in the inflammatory response.

View larger version (96K): [in this window] [in a new window]   Figure 4.   Ultrastructure of lungs exposed to hyperoxia. p21 (+/+) lungs (A and C) and p21 (/) lungs (B and D) were exposed to room air (A and B), or hyperoxia (C and D) for 84 h. Note the normal appearance of type II cells as assessed by abundant lamellar bodies and intact nuclei compared with the interstitial swelling, trapping of erythrocytes, and necrosis of endothelial and type I epithelial cells ( filled arrow). Extensive cell debris was observed in p21 (/) lungs exposed to hyperoxia (open arrow). Original magnification of sections: ×2,000.

View larger version (99K): [in this window] [in a new window]   Figure 5.   DNA fragmentation in lungs exposed to hyperoxia. p21 (+/+) (A, C, and E) and p21 (/) (B, D, and F ) mice were exposed to room air (A and B) or hyperoxia for 48 (C and D) or 84 (E and F  ) h. TUNEL staining was visualized as a brown stain. Sections were counterstained with methyl green. Filled arrows mark TUNEL-positive nuclei, open arrows mark TUNEL-negative nuclei; filled arrowheads mark TUNEL-positive cell debris. Sections reacted in the absence of terminal transferase had minimal TUNEL staining (data not shown, [4]). Bar = 100 µm.

DNA Fragmentation and Cell Death

Using mice obtained from separate exposures, we examined the lungs by electron microscopy in an effort to understand how p21 was protecting the lung. As expected, room air-exposed p21 (+/+) lungs were not morphologically distinct from room air-exposed p21 (/) lungs (Figures 4A, and 4B). Surprisingly, we also did not observe significant morphologic differences between the mice after 48 h of hyperoxia (data not shown). In these experiments we obtained several p21 (/) mice that survived to 84 h, which were used for comparison with p21 (+/+) mice that begin to show morphologic signs of injury by this time. The collective percentage of p21 (/) mice that survived to 84 h in all of our experiments was approximately 10%, during which time all p21 (+/+) mice remained viable. p21 (+/+) lungs exposed to hyperoxia for 84 h revealed early stages of interstitial edema among intact endothelial and type I epithelial cells (Figure 4C). In contrast, comparably exposed p21 (/) lungs showed extensive interstitial swelling with thickened appearance of alveolar septae that was associated with trapping of erythrocytes (Figure 4D). Closer examinations revealed marked swelling of mitochondria and nuclei from endothelial and type I epithelial cells associated with cell fragmentation. Nuclear fragmentation and other morphologic signs of apoptosis were not readily observed. These morphologic signs of necrosis have been previously described in wild-type mice and rats exposed to lethal levels of hyperoxia (1, 24). Although some mild swelling of terminal bronchiole and type II epithelial cells were noticed, these cells did not appear to be as injured as the alveolar endothelial and type I epithelial cells.

TUNEL staining was performed on p21 (+/+) and p21 (/) lungs to determine whether p21 deficiency modified DNA fragmentation during exposure to hyperoxia. Minimal TUNEL-positive staining was observed in p21 (+/+) and (/) lungs exposed to room air (Figures 5A and 5B). Abundant TUNEL-positive nuclei were detected among TUNEL-negative cells throughout the parenchyma of both mice after 48 h of hyperoxia (Figures 5C and 5D). TUNEL staining was still apparent in p21 (+/+) lungs exposed to hyperoxia for 84 h and was occasionally detected in alveolar cells that had separated from their basement membrane, consistent with the early signs of cell death (Figure 5E). TUNEL-negative cells were also occasionally detected. In contrast, abundant TUNEL-positive staining was observed in alveolar cell debris of the p21 (/) mice as well as in some intact nuclei (Figure 5F). Closer examination revealed that TUNEL-negative cells could also be found in the sections. Thus, DNA fragmentation was detected within intact nuclei of p21 (+/+) and p21 (/) lungs exposed to hyperoxia for 48 h. After longer exposures it was observed in cell debris of p21 (/) lungs.

Hyperoxia Inhibits DNA Replication through p21

The normal adult mammalian lung has a low rate of cell proliferation that decreases after 48 h of exposure to > 90% oxygen (20). Because decreased proliferation is observed when increased expression of p21 is detected (5), we wanted to determine whether p21 mediated the growth-arresting activities of hyperoxia. One technique used to assess proliferation has been to measure incorporation of [³H]thymidine or BrdU. Inasmuch as nucleotides are incorporated during DNA replication and DNA repair, it remains unclear how to interpret findings where both processes are likely to occur. The expression of PCNA has also been used to measure proliferation. This technique is limited because PCNA participates in DNA replication/ repair and has a long (20-h) half-life. We therefore chose to assess the expression of histone H3 mRNA as a measure of DNA replication because its expression is restricted to the S phase. Transcription of the H3 gene increases as cells enter the S phase and decreases when they enter the G2 phase. Because H3 mRNA is not polyadenylated, it decays rapidly when transcription ceases, due to the presence of newly synthesized histones. Because cells undergoing DNA repair do not synthesize histones, they also do not accumulate H3 mRNA. Perhaps most compelling is that its expression correlates with BrdU labeling in highly proliferative colonic epithelial cells by in situ hybridization (25).

p21 (+/+) and (/) mice were exposed to room air or hyperoxia for 24, 48, and 72 h, and H3 mRNA expression was analyzed by Northern blotting. As expected, H3 mRNA levels decreased in p21 (+/+) lungs during exposure to hyperoxia (Figure 6). In contrast, H3 expression remained unchanged in p21 (/) mice exposed to hyperoxia over the same time period. The large standard deviation at 24 h was due to one animal. Thus, hyperoxia decreases proliferation in p21-wild-type lungs but not in p21-deficient lungs.

View larger version (43K): [in this window] [in a new window]   Figure 6.   Hyperoxia inhibits histone H3 expression. (A) Total lung RNA from p21 (+/+) and p21 (/) mice exposed to room air (0) or > 95% oxygen for 24 to 72 h was hybridized for H3 and 18S RNAs. (B) The intensity of hybridization was quantified by PhosphorImage analysis and graphed relative to values obtained from p21 (+/+) mice exposed to room air. Hyperoxia significantly decreased H3 mRNA in p21 (+/+) mice at 48 and 72 h (n = 3 for all values; *P < 0.05) and did not alter its expression at any time in p21 (/) mice.

The population of cells expressing H3 mRNA was identified by in situ hybridization. H3 was detected in a few scattered cells in the parenchyma and terminal bronchioles of p21 (+/+) lungs exposed to room air (Figure 7A). Closer examination of the parenchyma revealed that the proliferating cells were most likely endothelial and type II epithelial cells, as previously reported (Figure 7E) (26). Similarly, H3-expressing cells were detected in terminal bronchioles and throughout the parenchyma of p21 (/) lungs exposed to room air (Figure 7B). p21 (+/+) lungs exposed to hyperoxia for 72 h had significantly fewer H3-positive cells, as expected when lungs are growth-inhibited by hyperoxia (Figure 7C). In contrast to the p21-wild-type lungs, H3-expressing cells were still detected in p21 (/) lungs after exposure to hyperoxia (Figure 7D). Examination under higher-power resolution showed that the proliferating cells were likely to be endothelial and type II cells (Figure 7F). Although H3 expression was detected in macrophages, they were not the predominant proliferating cell in the lung. Hybridization of antisense and sense H3 probes to the highly proliferative intestinal crypt cells demonstrated the specificity of the probe for proliferating cells (Figures 7E and 7F). Mice were also labeled with BrdU for 2 h, their lung cells were dissociated with proteases, and BrdU incorporation was assessed by flow cytometry. Using this technique, we confirmed that hyperoxia did not decrease proliferation in p21-deficient lungs (data not shown). Our observations suggest that the proliferating cells detected in p21 (/) lungs exposed to hyperoxia were likely to be the same cell types observed in the unexposed lung and not proliferation of new cell types such as inflammatory or bronchiole epithelial cells. We therefore conclude that hyperoxia inhibits pulmonary cell proliferation through induction of p21.

View larger version (123K): [in this window] [in a new window]   Figure 7.   Localization of H3 mRNA. p21 (+/+) mice (A, C, and E) and p21 (/) mice (B, D, and F ) were exposed to room air (A, B, and E) or hyperoxia (C, D, and F ) for 72 h. Sections from lungs were hybridized with [33P]-labeled antisense H3 riboprobe. Signal specificity was confirmed by hybridizing intestines with [33P]-labeled antisense (G) or sense (H) H3 probe. Sections were counterstained with H&E. E and F were imaged under higher power resolution than the other panels.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Previous studies have shown that hyperoxia inhibits proliferation, damages DNA, and kills microvascular endothelial and type I epithelial cells. It also increases p53 and p21, which are likely to mediate the cellular response to hyperoxia by inhibiting proliferation and promoting DNA repair. The present study extends these observations by demonstrating that hyperoxia increases p53-independent expression of p21 in murine lungs, and that p21 inhibits proliferation and enhances survival. One mechanism by which p21 may protect cells from hyperoxia is through its ability to restrict entry into the S phase during exposure to hyperoxia, thereby allowing additional time for cells to repair damaged DNA. Although our findings reveal that p21 inhibits proliferation in vivo, it has yet to be shown that hyperoxia damages DNA that is repaired through a p21-dependent process. Nevertheless, our findings demonstrate for the first time a role for p21 during oxidant lung injury and they underscore the potential importance of limiting DNA replication when DNA damage is occurring.

p21 exerts a G1 block through two distinct pathways. The amino-terminus binds to and inhibits the kinase activities of the G1 and S phase cyclins (18). The carboxy-terminus also inhibits DNA replication by binding PCNA. In this capacity, p21 plays an important role in blocking replication of proliferating cells, thereby preventing fixation of mutations in the genome. As shown in the late 1960s using tritiated thymidine, the normal adult murine lung has a relatively low mitotic index of approximately 2% that is thought to reflect renewal processes because the total population of pulmonary cells does not increase (26). Approximately 40% of the proliferating cells are leukocytes and 30% are microvascular endothelial cells. Type II epithelial cells comprise approximately 5% of the population with type I epithelial cells thought to be terminally senescent and unable to reproduce themselves. Because labeled endothelial cells decreased faster over time compared with other cell types, they have a higher mitotic index and therefore would be the most sensitive to genotoxins such as hyperoxia. In fact, lower levels of oxygen do inhibit proliferation and kill endothelial cells at concentrations that have no effect on epithelial cells (1, 20). Although type II epithelial cells have a slightly higher mitotic index than type I cells and express p21 during exposure to hyperoxia, we did not find that p21 deficiency affected their ability to express surfactant genes (data not shown) or survive in hyperoxia. p21 expression decreases during recovery in room air when type II cells proliferate and differentiate into type I cells. It remains to be determined whether p21 prevents premature replication of type II cells before repair has been completed. Nevertheless, our finding that necrosis of endothelial cells is associated with failure to inhibit proliferation is consistent with the concept that DNA replication is coupled with DNA repair.

Recent in vitro studies have confirmed our in vivo observation that hyperoxia inhibits DNA synthesis through p21-dependent mechanisms. Studies using simian virus 40- transformed alveolar type II epithelial cells demonstrated that hyperoxia inhibits their proliferation through both TGF--dependent signaling and induction of p21 that inhibited G1 cyclin E/Cdk2 kinase activity (27). Similarly, hyperoxia caused human bronchial smooth-muscle cells to growth-arrest in the G1 and S phases (28). Although p21 is likely to mediate the G1 arrest, it remains unknown whether it participates in the S phase accumulation that has been recently documented. Nevertheless, the present finding that p21 inhibits DNA replication in vivo is consistent with these in vitro studies.

p21 may also facilitate DNA repair independent of cell-cycle arrest through its interactions with PCNA. p21 promotes nucleotide excision repair by preventing binding of DNA endonuclease with PCNA (29). The p21-PCNA complex binds to sites of DNA damage and repair commences when p21 dissociates from PCNA. These observations are consistent with other studies demonstrating that DLD1 or HCT116 colon carcinoma cell lines deficient in p21 are sensitive to damage caused by cisplatin, nitrogen mustard, or UV irradiation (10, 30). Moreover, p21-expressing cells have a greater ability to repair in vitro-damaged reporter DNA than do p21-deficient cells. This activity was localized to the carboxy-terminal PCNA binding domain. In contrast, p21 does not protect cells from damage caused by adriamycin, ionizing radiation, taxol, or vincristine, which are likely to be repaired by base excision repair. This ability of p21 to repair damaged DNA may afford protection to the type I epithelial cell, which is considered to be terminally senescent. Based upon the importance of p21 in facilitating DNA repair, future studies using pulse-field gel electrophoresis, comet assays, and other DNA damage/repair assays should provide insight into DNA repair events. Because techniques to isolate pure populations of type I cells unfortunately do not exist, these studies will require appropriate model development. Additional studies using in vitro models are more likely to provide important information on how p21 protects pulmonary cells from hyperoxia that cannot be ascertained in vivo due to the cellular complexity of the whole lung.

It is also possible that p21 protects the lung independent of DNA replication and repair. Several studies have suggested that p21 may protect cells from genotoxins by preventing apoptosis. For example, adenoviral-mediated delivery of p21 into p21-deficient mouse embryo fibroblasts protected against p53-dependent apoptosis (31). Thus, loss of p21 may signal apoptosis through enhanced activity of other p53-dependent pathways such as Bax. Although hyperoxia increases Bax expression (3), morphologic signs of apoptosis have yet to be detected in vivo and the effect of hyperoxia on cell survival in the absence of Bax has yet to be determined. In contrast, the data support the concept that hyperoxia kills cells by necrosis because electron micrographs of rat and murine lungs exposed to hyperoxia revealed swelling of perinuclear and endoplasmic cisternae, mitochondria, and nuclei that was followed by cellular destruction (this study and Refs. 1 and 24). Similarly, the current study found that hyperoxia kills p21-deficient alveolar endothelial and type I epithelial cells that showed morphologic signs of necrosis. Moreover, A549 adenocarcinoma cells exposed to hyperoxia die by necrosis (32). On the basis of these findings, it is likely that the sensitivity of p21-deficient mice to hyperoxia is due to an acceleration in the rate of alveolar cell necrosis and not through changes in the manner in which cells die.

A recent study using cDNA microarray technology demonstrated that regulated expression of p21 in HT1080 human fibrosarcoma cells modified the expression of many genes that are unrelated to cell-cycle progression and DNA repair (33). In this study, the authors compared the response of cells to regulated expression of p21 and simple serum starvation, which also promotes G1 growth arrest. After subtracting for genes that are induced by serum starvation, a high correlation between p21 expression and increased expression of extracellular matrix genes, such as fibronectin, plasminogen activator inhibitor-1, and integrin 3 was observed. p21 also increased the expression of several lysosomal and mitochondrial enzymes. The finding that many of these p21-induced genes are also expressed by senescent cells is interesting because it suggests that p21 may protect senescent type I epithelial cells through mechanisms independent of DNA replication and repair. As mentioned earlier, this hypothesis can be tested once appropriate cell-line models have been established.

On the basis of the observation that p21 protects the lung from hyperoxic injury, it will be important to clarify how ROS induce p21 in vivo. p21 mRNA may be stabilized post-transcriptionally by oxidative stress through the mitogen-activated protein kinase pathway (17). This finding may explain why p53-independent induction of p21 was not observed in p53-deficient lungs exposed to 92% oxygen (6) whereas modest induction was observed at higher oxygen levels (the present study). p21 is also induced in keratinocytes by TGF- and in osteoblasts by IL-6-type cytokines (14, 15). TGF- and IL-6 expression increases in lungs exposed to hyperoxia (22, 34). It remains to be determined whether they are responsible for the induction of p21 observed in the p53-deficient mice. Because transgenic mice overexpressing IL-11 in the lung are highly resistant to hyperoxic injury and death, it would be interesting to examine the interactions of IL-6-type cytokines (such as IL-11) and p21 during hyperoxic lung injury (7).

In summary, the present study demonstrates that ROS produced by exposure to hyperoxia induces p53-independent expression of p21 in the lung. The absence of p21 results in rapid necrotic alveolar cell death and mortality associated with an inability to inhibit DNA replication. To our knowledge, this is the first report demonstrating functional relevance of members of the p53 suppressor pathway in oxidant lung injury. Our findings reveal that induced expression of p21 protects against oxidant injury and suggest that it could be used as a novel therapeutic agent for patients exposed to supplemental oxygen as well as other types of inhaled oxidant gases. Moreover, because p53 and p21 are induced in fibrosis, such as bleomycin-induced fibrosis, it will be exciting to examine their role in other types of lung diseases where cell injury and altered proliferation are observed (35).