Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Pathophysiologic pathways leading to chronic lung disease of prematurity (CLD) are believed to include mechanical cellular injury and inflammation in addition to oxidant stress (1, 2). Inflammatory cell influx, particularly of neutrophils, can contribute to oxidant stress and cellular injury through generation of reactive-oxygen species (ROS) (3). Impaired alveolar development characterizes modern-day CLD following oxidant stress and inflammation, which is also seen in large-animal models of CLD designed to mimic neonatal intensive care practices (4, 5), as well as in neonatal hyperoxia-exposure experimental models of CLD (6).

We and others have shown that hyperoxia exposure leads to increased neutrophil chemokine expression (9, 10). In hyperoxia-exposed newborn rats we have shown that treatment with neutralizing antibodies to neutrophil chemokines can reduce lung inflammation (9). More recently we determined that blocking neutrophil influx during hyperoxia exposure preserved normal alveolar development. This may be attributed to effects of preservation of cell proliferation and/or DNA repair in newborn rat lung: proliferating cell nuclear antigen expression and bromodeoxyuridine uptake were preserved (8). We do not yet know how reducing neutrophil influx attenuates hyperoxia effects on newborn lung cell proliferation and alveolar development.

Appropriate cell proliferation following oxidative DNA damage depends on successful DNA repair, which in turn depends on intact nuclear signal transduction and protein synthesis (11). Hyperoxia has been shown to increase DNA damage as measured by terminal transferase nick end- labeling (TUNEL) in adult rats and mice (12), which is attenuated in transgenic mice overexpressing counterinflammatory cytokines such as interleukin-11 (15). These observations were initially interpreted as changes in apoptosis, but further ultrastructural analysis in hyperoxia-exposure studies indicate that few of the TUNEL positive cells in hyperoxia-exposed adult mouse lung are in fact apoptotic (14). O'Reilly and colleagues have suggested that DNA nicking during oxidative stress more adequately explains this phenomenon (11, 16). Whether apoptosis contributes significantly to abnormal alveolar development in CLD is unknown.

Neutrophil influx accompanies hyperoxic stress in the newborn, and neutrophils are a significant source of ROS, likely contributing to the total oxidant burden (3). We therefore hypothesized that directly reducing neutrophil influx during hyperoxia exposure by neutralizing neutrophil chemokine would decrease nucleic acid damage (17) and possibly apoptosis, because neutrophil-mediated paracrine factors activate the Fas cell death pathway in vitro in pulmonary epithelial cells (18).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Hyperoxia Exposure

Newborn rats were exposed to 95% O₂ and 5% air, or air alone, beginning on the day of birth, within 10 h of delivery, as previously described in detail by Bruce and colleagues, and continuing until the rats were killed 8 d later (19). Rat pups from four litters were randomly assorted into four recombined litters, and were exposed to 95%O₂ or air in sealed Plexiglas cages 60 × 45 × 25 cm, which we have previously described in detail (8, 20). On Days 3 and 4, hyperoxia-exposed pups were injected intraperitoneally with 10 µg (~ 2 mg/kg body weight) purified goat antirat cytokine-induced neutrophil chemoattractant (CINC; R&D Systems, Minneapolis, MN) or nonimmune goat immunoglobulin G (IgG) (Sigma, St. Louis, MO). We previously found this to be the optimal dose to block neutrophil influx (8). Air-exposed animals were not injected.

At 8 d, pups were killed with sodium pentobarbital 150 mg/kg intraperitoneally, and a tracheal cannula was placed. An abdominal incision was made, and the diaphragm was punctured carefully to collapse the lungs. Lungs from four animals in each group were inflation-fixed via tracheal cannula at 20 cm H₂O for 30 min using 10% phosphate-buffered formalin. After overnight immersion in fixative, the lungs were embedded in paraffin, and completely sagitally sectioned at 5 µm thickness.

TUNEL

Random sections were obtained from the middle third of the serial sections, from four animals per treatment group. Two sections from the lungs of each pup were dewaxed, rehydrated, and labeled with rhodamine-deoxyuridine triphosphate and terminal deoxytransferase according to the manufacturer's directions (In Situ Cell Death Detection Kit; Roche Molecular Biochemicals, Indianapolis, IN). Additional sections were labeled after pretreatment with DNase I (DNase I) (Roche Molecular Biochemicals) for 30 min at 37°C to induce universal DNA nicking (positive control) or were incubated without terminal transferase (negative control). Sections were counterstained with 4',6 diamidino-2-phenylindole (DAPI) containing mounting medium diluted 1:3 in DAPI-free medium (Vectashield; Vector Labs, Burlingame, CA). Four 200× microscopic fields per section were digitally imaged with a Nikon Diaphot inverted fluorescence microscope using a UV-light source and a digital camera (CoolPix 990; Nikon, Mellville, NY). Images were obtained at emission wavelengths of 590 (rhodamine: TUNEL positive) and 380 nm (DAPI). Fluorescence intensities for TUNEL and DAPI signal were quantified using density threshold image analysis of the digital images using NIH Image software (version 1.62). Rhodamine signal (TUNEL positive DNA) was normalized to DAPI signal (total DNA) for each captured photomicrographic field. Mean labeling indices for each treatment group were expressed as a proportion of the mean normalized TUNEL signal in the positive control group (maximum). Results for each condition are mean observed  maximum TUNEL (DNase I) from four animals per treatment group.

Immunohistochemical Measurement of Nucleic Acid Oxidation: 8-OH-2'-Deoxyguanosine

Random sections were obtained as noted above. After dewaxing and rehydrating, antigen retrieval was performed by microwaving sections immersed in 0.1 M Na-citrate buffer, pH 6, in a glass Coplin dish for 5 min at 900 W. After cooling in the buffer, sections were rinsed in water and incubated in 2N HCl at 37°C for 30 min, rinsed in phosphate-buffered saline (PBS) + 0.1% Tween80 (PBS-T) (Sigma), then blocked in 10% horse serum in PBS for 1 h at room temperature. Sections were incubated with monoclonal mouse anti-8-OH-2'-deoxyguanosine-(OHdG) 1:50 (Oxis International, Portland, OR) in 2% horse serum:PBS overnight at 4°C. The monoclonal antibody has previously been determined not to crossreact with 8-OH guanosine or 8-OH-guanine (21).

After three PBS-T washes the sections were incubated with 1:500 antimouse biotinylated IgG in PBS-T for 1 h at room temperature. This was followed with one 10-min PBS-T wash, at which point the sections were treated with 0.3% H₂O₂/methanol for 30 min to quench endogenous peroxidases. Sections were washed twice with PBS-T and incubated with Vector ABC-Elite (Vector Labs) for 45 min. The OHdG was detected with 3-3'-diaminobenzidine (DAB) substrate (DAB Kit; Vector) and counterstained with hematoxylin.

Negative controls for signal specificity included the following: treatment with DNase I 100 µg/ml at 37°C for 4 h, followed by washing in four changes of PBS, 15 min each, to reduce DNA attributable signal, and competition during primary antibody incubation with 100-fold molar excess OHdG (Sigma).

After counterstaining with hematoxylin and mounting, four random nonoverlapping 400× photomicrographic images were obtained from two random sections from each animal using a digital camera as above. A labeling index was obtained by counting OHdG-positive cells and dividing by the total number of nuclei per high-power field. Because we noticed that the variation of OHdG labeling was decreased in bronchiolar epithelium, we counted bronchiolar and alveolar epithelial cells separately. At least 500 cells per animal were counted from each of four animals per treatment group.

Colocalization of OHdG and TUNEL

OHdG and TUNEL detection were performed as noted above, in sequence, with the following modifications. The HCl incubation was omitted following the microwave step before proceeding with OHdG detection which was done as described. Following DAB color development, slides were blocked in 0.1M Tris-HCl, 3% bovine serum albumin, and 20% bovine serum for 30 min at room temperature. TUNEL was then detected according to the manufacturer's directions, which differed from the previous method by using a fluorescein-labeled dUTP in place of a rhodamine-labeled dUTP, which was subsequently detected with anti-fluorescein:horseradish peroxidase conjugate (In Situ Cell Death Detection-POD; Roche). In place of DAB, a substrate that produces a violet precipitate was used (VIP Substrate; Vector). Representative photomicrographs were obtained as noted above.

Immunohistochemical Markers of Apoptosis

To determine whether widespread hyperoxia-induced TUNEL signal was attributable to apoptosis, we performed immunohistochemical measurements of Bax, a proapoptotic mitochondrial proto-oncogene product, and Bcl-2, an antiapoptotic proto-oncogne product (22, 23). Adjacent sections were labeled for Bax and Bcl-2 to evaluate the relative abundance in similar cell types, because it is the ratio of Bax and Bcl-2 that determines progression to apoptosis (24). Sections were dewaxed as before. The Bcl-2 sections were treated for antigen retrieval as described previously. The Bax sections were not, because this was previously determined not to affect sensitivity. Both were treated with 0.3% H₂O₂/ methanol to quench endogenous peroxidases and subsequently blocked in 10% goat serum/PBS. Sections were then incubated overnight at 4°C in polyclonal mouse anti-Bcl-2 (BD PharMingen, San Diego, CA) 1:1,500, or polyclonal mouse anti-Bax 1:100 dilution (BD PharMingen). After washing with PBS-T, anti-Bcl-2-treated sections were incubated with a secondary biotinylated antirabbit IgG (Vector) at 1:1,000 for 1 h and detected with Vector ABC Elite according the manufacturer's directions, using DAB, and were counterstained with hematoxylin. Anti-Bax treated sections were incubated with a monoclonal antirabbit IgG:peroxidase (Sigma) at 1:500 for 1 h, and detected with DAB as above.

Positive and negative controls for apoptosis were obtained by culturing A549 epithelial cells in Ham's F12K+10% fetal bovine serum, 75 U/ml penicillin G, 75 µg/ml streptomycin (Life Technologies, Rockville, MD), at 37°C in 95% air, 5% CO₂ until 30-40% confluence was attained. Apoptosis was induced by replacing the media without serum, and continuing to culture 72 h (25). Controls had media replaced containing serum. Cells were then treated with trypsin for 5 min., rinsed in fresh media + serum, and cytocentrifuged before fixing in methanol for 5 min. Bax and Bcl-2 were detected as above, except the antigen retrieval step was omitted.

To detect cells committed to apoptosis we performed histochemical detection of M30, an epitope produced by caspase 6 digestion of cytokeratin 18, an intermediate filament in lung epithelium: caspase 6 activation is necessary for cell commitment to apoptosis (26, 27). We used a mouse monoclonal antibody, anti-M30, according to the manufacturer's directions (Roche Molecular Biochemicals). Sections were dewaxed, rehydrated, and treated in a 0.1% trypsin 20 mM TRIS pH 8, CaCl₂ 0.1% at 37°C for 15 min. Sections were blocked in 1% BSA in PBS-T. Sections were incubated with anti-M30 1:10 overnight in the blocking buffer, washed in PBS-T, and followed by incubation with biotinylated antimouse IgG (Vector) at 1:100 in PBS-T for 1 h, washing in PBS-T, then detection with Vector ABC-Elite. Color was developed DAB and counterstained as before. Control and apoptotic A549 cells were used as assay controls for M30 detection of apoptosis and detected as above, omitting the trypsin digestion.

Between groups, differences were tested by analysis of variance, and post hoc analyses were performed by Kruskal-Wallis-Kramer using JMP software (SAS, Cary, NC) (28, 29). Significance was accepted at P < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

We previously demonstrated that anti-CINC treatment at 3 and 4 d prevented 95% O₂-induced neutrophil accumulation at 8 d in bronchoalveolar lavage and in histologic sections from the animals studied in this report (8).

DNA Nicking: TUNEL

95% O₂ exposure for 8 d induced widespread TUNEL signal in newborn rat lung (Figure 1). Blocking neutrophil influx by treatment with anti-CINC 10 µg at 3 and 4 d prevented the hyperoxia-induced DNA nicking as measured by TUNEL, reducing TUNEL signal to air control levels (Figures 1 and 2). Maximum rhodamine (TUNEL) fluorescence was seen in the DNase I-treated sections. DAPI fluorescence appeared unaffected by DNase I treatment in these conditions, and omission of terminal transferase eliminated TUNEL signal without affecting DAPI signal (data not shown). Brightfield examination of these sections at 400× and 1,000× rarely revealed apoptotic bodies, although this was not strictly quantified.

View larger version (66K): [in this window] [in a new window]   Figure 1.   Effect of anti-CINC on 95% O2 × 8 d-induced fluorescent TUNEL signal in newborn rat lung. Sections were obtained from air (A), 95% O2 + IgG (B), or 95%O2 + anti-CINC-treated animals (C). Maximal signal induced by DNase I treatment (D). Insets are total nuclear (DAPI) signal. Original magnification: ×400.

View larger version (26K): [in this window] [in a new window]   Figure 2.   Effect of 95% O2 × 8 d ± anti-CINC on TUNEL fluorescence (TUNEL/DAPI), normalized to DNase I (maximum signal). Mean ± SEM. *P < 0.05 versus 95% O2 + IgG control group. n = 4/group.

DNA Oxidation: OHdG Immunohistochemistry

95% O₂ exposure for 8 d induced widespread DNA oxidation in newborn rat lung (Figure 3). Air-exposed animals demonstrated little OHdG signal, largely confined to bronchiolar epithelium. Semiquantitative analysis showed that treatment with anti-CINC 10 µg at 3 and 4 d significantly (P < 0.05) prevented hyperoxia-induced DNA oxidation in alveolar epithelium (Figure 4). There was a trend toward reduction of OHdG in bronchiolar epithelium, although this did not reach statistical significance. Pretreatment of sections from 95% O₂ + IgG-treated animals with DNase I or incubation with excess OHdG reduced or eliminated signal (Figure 3, inset), demonstrating specificity of the immunostaining.

View larger version (44K): [in this window] [in a new window]   Figure 3.   Effect of 95% O2 × 8 d ± anti-CINC on oxidized DNA (OHdG immunohistochemical signal) in newborn rat lung. OHdG label in alveolar epithelium after air exposure (A), after 95% O2 exposure (B), after 95% O2 exposure + coincubation with OHdG (negative control, B inset), after 95% O2 exposure + anti-CINC treatment (C); bar = 50 µm. Colocalization of OHdG and TUNEL (violet) in bronchiolar (D) and alveolar (E) epithelium; bar = 50 µm.

View larger version (23K): [in this window] [in a new window]   Figure 4.   Effect of 95% O2 × 8 d ± anti-CINC on OHdG labeling index. Relative abundance of OHdG in alveolar (gray) and bronchiolar (black) epithelial cells. Mean ± SEM. *P < 0.05 versus 95% O2 + IgG group. n = 4/group.

Colocalization of TUNEL and OHdG

OHdG staining, both nuclear and cytoplasmic, could easily be seen in both bronchiolar (Figure 3D) and alveolar (Figure 3E) epithelium. TUNEL signal was confined to nuclei.

Bax/Bcl-2 Immunohistochemistry

To determine whether the observed widespread TUNEL signal was attributable to apoptosis, we screened random sections obtained as above for substanial increases in Bax (proapoptotic) or decreases in Bcl-2 (antiapoptotic) signal. As shown in Figure 5, 95% O₂ exposure for 8 d modestly increased both Bax and Bcl-2 in alveolar and bronchiolar epithelium. Anti-CINC treatment had no obvious effect on either Bax or Bcl-2 immunostaining in lung epithelium, whereas serum deprivation induced near universal Bax signal and eliminated Bcl-2 signal in A549 cells.

View larger version (116K): [in this window] [in a new window]   Figure 5.   Effect of 95% O2 × 8 d ± anti-CINC on Bax (A, C, E) and Bcl-2 (B, D, F ) immunohistochemical labeling in newborn rat lung. Sections were obatined from air (A, B), 95% O2 + IgG (C, D), or 95% O2 + anti-CINC-treated animals (E, F ). Bax (G, H) and Bcl-2 (I, J ) signal in serum-fed (apoptosis-negative control, G, I ) and serum-deprived A549 cells (apoptosis-positive control, H, J). Bar = 50 µm.

M30 Immunohistochemistry

We found very little M30 immunostaining in lung parenchyma in any treatment group, as shown in representative low-magnification photomicrographs (Figure 6). Detailed high-power microscopy showed positive staining in distal alveolar epithelium in some cases (Figure 6, inset). Positive cells were more easily detected in sections from 95% O₂ + IgG-exposed rat lung than in air or 95% O₂ + anti-CINC treatment groups, but these changes were not quantified. Compared with TUNEL and OHdG labeling, M30 immunostaining of lung epithelium was infrequent. A549 cells cultured without serum (positive apoptosis control) were typically M30 positive in contrast to cells cultured with serum (negative apoptosis control), as shown in Figure 6. Cells detected with IgG alone showed no labeling (not shown).

View larger version (145K): [in this window] [in a new window]   Figure 6.   Effect of 95% O2 × 8 d ± anti-CINC on M30 (indicator of caspase 6 activity on cytokeratin 18) in newborn rat lung. Sections were obtained from air (A), 95% O2 + IgG (B), or 95% O2 + anti-CINC-treated animals (C). M30 signal in serum-fed (apoptosis-negative control, D) and serum-deprived A549 cells (apoptosis-positive control, E). Bar = 100 µm. Inset is high-power detail, bar = 50 µm.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Oxidative biomolecular damage among other mechanisms is believed to contribute to CLD, as recently reviewed by Frank and Sosenko (30). Neutrophils can contribute to oxidative DNA damage, as demonstrated in vitro by Knaapen and colleagues (31), and could potentially contribute to lung cell damage and death through other mechanisms, such as elaboration of Fas ligand (18) or proteases (32).

Our results show that 95% O₂ exposure for 8 d induces widespread DNA nicking (TUNEL) and oxidation (OHdG), and that blocking neutrophil influx significantly reduces DNA strand breaks and nucleic acid oxidation in lung parenchymal cells. We conclude that neutrophil-generated ROS contribute significantly to nucleic acid damage in newborn rat lung even during very high ambient oxygen exposure. Reducing oxidative DNA damage in newborn lung subjected to significant oxidative stress may therefore require manipulation of the inflammatory response. Whether this is the case at lower degrees of oxidative stress is as yet unknown.

The magnitude of hyperoxia-induced TUNEL and OHdG labeling was similar, so we suspected that most of the increased TUNEL signal was in part due to oxidative damage rather than predominantly due to apoptosis, because widespread apoptosis was not readily observed morphologically in our model. The prevention of TUNEL and OHdG signal accumulation in rats treated with anti-CINC treatment was likewise comparable. Although TUNEL has been used to assess apoptosis in several investigations evaluating hyperoxia-induced lung injury in adult animal models (12, 15, 33), and recently, in newborn mice (34), it is not specific to apoptosis (14). We therefore sought to determine whether the widespread DNA damage we detected with hyperoxia and ameliorated with neutrophil blockade was substantially attributable to apoptosis.

In contrast to the large hyperoxia-induced increase in TUNEL and OHdG label in lung epithelium, we found only modest hyperoxia-induced decreases in Bcl-2 (antiapoptotic) or increases in Bax (proapoptotic) expression as determined by immunohistochemistry. Treatment with anti-CINC had no obvious effect on in vivo Bax or Bcl-2 expression. There was no apparent effect of hyperoxia or anti-CINC on the relationship of Bax and Bcl-2: during apoptosis, Bax expression increases relative to Bcl-2 (24).

Our findings agree with those in adult hyperoxia-exposed mice reported by O'Reilly and colleagues (14). Because cells undergoing apoptosis in response to injury would not be expected to exhibit markers of apoptotic steps synchronously, we also performed histochemical measurement of M30, which is manifested only when caspase 6 has sufficiently digested epithelial intermediate filament protein cytokeratin 18 to expose the neo-epitope. This step in the apoptosis cascade precedes internucleosomal DNA fragmentation that would manifest as TUNEL (35). M30-positive cells were present in the 95% O₂-treated group, but were less frequently seen than TUNEL, OHdG, or Bax-positive cells in hyperoxia-exposed lung. Pulmonary vascular smooth muscle cells, which undergo widespread apoptosis during early postnatal remodeling (36), were typically M30 positive.

We were looking for widespread effects on apoptosis, because the changes in TUNEL and OHdG during hyperoxia exposure were large. We cannot exclude a role for apoptosis in this model of CLD, which almost certainly occurs in vivo. Apoptosis has been demonstrated in vitro in primary cell culture of hyperoxia-exposed newborn murine lung cells (34). In our studies it appears that the prevalence of apoptotic cells following hyperoxia exposure in vivo is considerably smaller than the prevalence of DNA nicking and DNA oxidation at the stage of injury we evaluated. These findings agree with Kazzaz and colleagues, who concluded that severe hyperoxia does not cause cell death predominantly by apoptosis in adult animals (12). Apoptosis may be more prevalent during later remodeling in response to injury. In an autopsy study of ventilated premature newborns with CLD, apoptosis, detected by a combination of TUNEL and morphologic assessment, was uncommon in acute (stage I by histologic criteria) CLD, and more common in higher severity CLD, which is consistent with our present findings in the acute hyperoxia- exposed newborn rat (37). The hyperoxia induced TUNEL signal we observed in our model may represent oxidative damage as suggested by the OHdG immunostaining pattern and magnitude, and the nuclear colocalization of TUNEL and OHdG signal that we demonstrated would support this premise.

DNA damage and repair may therefore be mechanisms more important than apoptosis in contributing to hyperoxia-impaired newborn lung development. We and others have reported that hyperoxia depresses proliferative/ repair capacity as measured by its effects on proliferating cell nuclear antigen (PCNA) expression and bromodeoxyuridine (BrdU) uptake (8, 37). We previously reported that blocking neutrophil influx restored hyperoxia-induced suppression of PCNA expression, and BrdU uptake using the same model of hyperoxia exposure and anti-CINC treatment (8). Our findings that DNA strand breaks and nucleic acid oxidation are both ameliorated by neutrophil blockade therefore support the general scheme of oxidant cellular injury inducing neutrophil influx, which in turn further contributes to ROS exposure and further oxidant damage. DNA damage may then obligate cells to arrest or delay their cell cycling leading to mitosis (11).

Neutrophil influx may obviously influence pulmonary cell injury and survival through nonoxidant mechanisms that we did not evaluate. Neutrophil elastase has been implicated in a number of acute lung injury models, and decreased antiprotease expression (relative to levels in healthy adults) has been suggested as an injury mechanism in premature newborns that develop CLD (32). Neutrophils may directly influence epithelial survival via Fas- dependent effects on pulmonary epithelial cell death (18, 38).

In summary, we found that blocking neutrophil influx during severe oxidant stress markedly reduced hyperoxia-induced DNA nicking and oxidation in newborn rat lung, and that apoptosis played a relatively minor role in the acute phase of this model of CLD. We conclude that neutrophil influx is a significant source of pulmonary oxidative DNA damage in newborns even at very high ambient inspired oxygen. We speculate that strategies aimed at blocking neutrophil influx or function in newborns at risk to develop CLD will avoid DNA damage and its adverse effects on cell proliferation and alveolar development.