Main

BPD is a lung disorder which is associated with the need for prolonged respiratory assistance and hyperoxia in low birth weight infants. Although the severity of BPD has moderated since first described by Northway et al.(1), it remains a significant problem in those newborns affected. Not only is there increased acute morbidity and mortality associated with BPD but also there are long-term sequelae. Among these sequelae are retarded alveolar development, a predisposition to respiratory tract infections, abnormal pulmonary mechanics, and airway hyperreactivity(2–4). Although the causative agent for BPD has not been conclusively identified, the hyperoxia required by affected infants is believed to be a major factor. Thus, the effect of hyperoxia on the newborn lung has been the subject of extensive studies. With the exception of the premature baboon model of BPD(5), the majority of these investigations have studied the effect of hyperoxia on neonatal animals, mostly rats. Neonatal rats were found to be virtually resistant to the lethal effects of hyperoxia, whereas sexually mature rats experienced 100% mortality(6). This difference was attributed to induction of antioxidant enzyme activity in the lungs of the neonatal rats, a response not seen in the adult rats(7, 8). This antioxidant enzyme effect has led to the current paradigm that ROS are directly responsible for the toxicity of oxygen to the lung. Despite resistance to lethality, exposure of neonatal rats to >97% oxygen did result in a reduction in total lung volume and ventilatory unit volume(9). Additionally, there was a reduction in length and an alteration of the structure of lung elastic fibers(9).

A number of in vivo and in vitro interventions have proven successful in mitigating the effects of hyperoxia on the neonatal animal lung. Epidermal growth factor increased the antioxidant enzyme and surfactant system in lung slices and reduced histopathologic damage associated with growth in >90% oxygen(10). Frank and his group reported that prenatal dexamethasone(11) and an antenatal propylthiouracil or a 21-aminosteroid protected neonatal rats from the adverse histopathologic effect of >95% oxygen(12, 13).

Resnick et al.(14) presented an alternate approach to the study of the effect of oxygen on the developing lung. They reported that exposure of fetal lung explants from a 12-d gestation mice to>95% oxygen resulted in complete inhibition of branching morphogenesis. We were interested in whether the 12-d gestation mouse fetal lung explant could be used to investigate the mechanism by which hyperoxia produced damage in the developing lung. The advantage of this preparation over either the use of lung slices from later gestation fetuses or neonatal animals is that the 12-d gestation lung explant undergoes branching morphogenesis, which is quantifiable microscopically. This explant preparation has the potential to be useful in the evaluation of compounds whose purpose is to reduce the effect of hyperoxia on the developing lung. The obvious disadvantage is that the stage of lung development in the 12-14-d gestation mouse fetus is substantially earlier than an early third trimester human infant.

Thus, we set out to determine the dose-response effect of oxygen on branching morphogenesis in mouse fetal lung explants. Our goal was to establish an oxygen concentration where any reduction of oxygen toxicity to the explant could be quantified. Then, to find out if ROS were involved in the effect on branching morphogenesis, we planned to add a variety of antioxidants targeted to specific ROS and to determine whether any of them could reduce the adverse effect of hyperoxia on explant branching morphogenesis.

METHODS

Animals. Timed pregnant CD-1 mice were obtained from Charles River Laboratories and were sacrificed by intraperitoneal pentobarbital, 100 mg/kg, on gestation d 12 or 14. The embryos were removed, and the lungs were dissected and divided into right and left lobes using a dissecting microscope. The separated lung lobes were initially placed in cold DMEM (Sigma Chemical), and within 4 h only the right lobes were transferred to membrane filter disks(Anodisk, Whatman Scientific, Clifton, NJ).

Incubation. The membrane filter disks were then placed in 60× 15-mm tissue culture dishes (Falcon 3037) containing 4 mL of DMEM in the center well. A 48-h incubation period was chosen because by 72 h, the airways began to swell and branching morphogenesis slowed, suggesting reduction of explant viability. Growth in normoxia during the first 24 h after removal from the fetus was minimal; however, between 24 and 48 h of incubation, bud and branch growth greatly accelerated. For the thymidine incorporation studies, 0.5 mCi/ml [methyl-3H]thymidine was added to the medium at the beginning of the incubation and remained there for the 48-h incubation. Viability was assessed on collagenase-digested explants after 48 h of incubation in 95%/5% (O2:CO2) by the ability of the isolated cells to exclude trypan blue.

Incubation. Culture dishes containing the lungs and DMEM were placed on a platform shaker (Labquake Shaker, Labindustries, Berkeley, Ca) inside a sealed 37°C water bath for 48 h with a gas flow of 7L/min. Fresh medium or medium containing the compound to be tested was added every 24 h. The incubation gas consisted of either air or O2:N2 (95%:0%, 75%:20%, 50%:45%, 35%:60%, or 10%:85%) with 5% CO2. The concentration in the incubator was constantly monitored using a Servomex 570 oxygen analyzer(Servomex, Servomex Ltd., Crowborough, Sussex, England). The role of the different ROS in producing the oxygen toxicity was evaluated by incubating the lungs in the presence of a variety of antioxidants and radical scavengers(Table 1): the general reductant,N-acetylcysteine combined with the synthetic glutathione peroxidase, ebselen (provided through the courtesy of Rhone-Poulenc Rorer, Koln, FRG); the nitric oxide synthase inhibitor, L-NAME; a synthetic vitamin E, Trolox,(provided through the courtesy of Dr. H. Bhagavan, Hoffman LaRoche, Nutley, NJ); a xanthine oxidase inhibitor, allopurinol; the iron chelator, deferoxamine; a synthetic superoxide dismutase mimetic, TMPYP (provided through the courtesy of Drs. Irwin Fridovich and Brian Day, Duke University Medical Center, Durham, NC); MnSOD, CuZnSOD; catalase; and finally, the combination of catalase and CuZnSOD. Unless stated otherwise, all reagents were purchased from Sigma.

Table 1 Antioxidants and inhibitors of reactive oxygen species formation tested
Full size table

Photomicrography and counting procedure. At the beginning and end of the 48-h exposure, lungs on the filter disk were photographed ×16 magnification using an Olympus Vanox photomicroscope with a 4× Plan Apo objective. The before and after photographs were compared for changes in the number of buds and branches. A branch was defined as an airway with two or more buds. For ease in statistical analyses, after 48 h of incubation a value of 1 (rather than 0) was chosen to indicate no increase from the number of buds and branches present after the initial isolation. A value of 2 meant that there was an increase of one branch (or bud) compared to the number immediately after the initial isolation (at time 0).

[methyl-3H]Thymidine counting. At the end of 48 h of incubation, filter disks containing the lungs incubated with the thymidine were washed five times with fresh DMEM. (A preliminary study showed that there was no residual radioactivity remaining after five washings.) The lungs were then washed from the disk with 1.2 mL of 0.05 M potassium phosphate buffer, pH 7.0, with 0.1 M EDTA, into a 1.5-mL Eppendorf tube and centrifuged for 7 min(12,500 × g). The supernatant was removed, and 550 μL of buffer were added to the lung pellet. The pellet was sonicated (Bronson Sonic Power Co., Danbury, CT) for 30 s × 5 and then recentrifuged as before. The supernatant was transferred to a clean 2-mL centrifuge tube; 250 μL were then added to 2.75 mL of scintillation fluid and counted for 20 s on a liquid scintillation counter (1211 Rackbeta, LKB, Wallac, Turku, Finland). DNA was determined according to the method of Gold and Souchat(15).

Statistics. Data are expressed as mean ± SE. Bud and branch development was determined on the same lobe for all comparison experiments. Statistical significance was determined using ANOVA and was defined as p < 0.05. If significant by ANOVA, the source of the significance was determined using ANOVA for two variables only.

RESULTS

Exposure of the fetal lung explants to different concentrations of oxygen ranging from 10 to 95% yielded surprising results. There was a significant increase in both bud and branch growth in explants after 48 h of growth in 21% oxygen (Fig. 1,A andB). Under hypoxic conditions (10% oxygen), growth could not be quantified after 48 h because the explants lost identifiable morphology (buds and branches could not be counted). Exposure to oxygen concentrations of 75% or greater for 48 h also resulted in complete inhibition of growth (Fig. 2,A andB) and partial loss of morphology. At 50% exposures a small amount of growth occurred(Fig. 3,A andB). Growth inhibition at 50% oxygen and greater was irreversible which was demonstrated by the failure of lungs incubated for 24 h at 50% oxygen to grow when 21% oxygen was restored for an additional 48 h.

Figure 1
figure 1

Appearance of left lobe from 12-d gestation mouse immediately after removal from the fetus (A). Growth of explant at 37°C after 48 h of incubation in 21% oxygen (B). Branches increased from 1 to 3 and buds from 4 to 10. Explants were grown in conditions described in “Methods.”

Full size image
Figure 2
figure 2

Representative example of appearance of lobe explant immediately after removal (A) from the 12-d gestation mouse fetus and after 48 h of growth in 75% oxygen (B).

Full size image
Figure 3
figure 3

Effect of 50% oxygen on the fetal lung growth. Appearance immediately after isolation and before incubation (A) and after 48 h of growth in 50% oxygen (B). No growth can be detected in the number of new buds or branches.

Full size image

Figure 4A shows that compared with 21% oxygen, all of the other oxygen exposures showed a significant reduction in the number of new branches formed, p = 0.0004 or less. There was significantly more growth for 35% exposures compared with 50%, p = 0.02. There was also significantly more branch formation at 50% compared with 75, 95, and 10% which can be attributed to the fact that there was no growth at those other oxygen exposures. However, when the formation of new buds was analyzed, there was no significant difference between the 21 and 35% exposures (Fig. 4B). There were significantly more buds formed at 21 and 35% oxygen than any of the other exposures, p = 0.002 or less. As with the branch data, there was significantly more bud formation at 50% oxygen than 10, 75, or 95% oxygen due to the fact that there was virtually no additional bud formation at those concentrations. Viability after 48 h of incubation was assessed in the 95% oxygen exposed explants. Collagenase-separated cells were viable as assessed by their ability to exclude trypan blue after exposure to 48 h of 95% oxygen. This provides evidence that, although hyperoxia prevents branch and bud development, it does not appear to be lethal to the explant.

Figure 4
figure 4

Increase in number of branches (A), and buds(B) after 48 h of incubation in the different concentrations of oxygen. (A) * = significantly greater than the other oxygen exposures (p < 0.0004); ** = significantly greater than 50%(p = 0.02), and the 10, 75, and 95% concentrations (p = 0.0001); t = significantly greater than the 10, 75, and 95% concentrations (p = 0.0001). (B) * = significantly greater than all the other oxygen concentrations (p = 0.0001) except 35%(p = 0.11); ** = significantly greater than the 10, 75, and 95% concentrations (p = 0.0001) and 50% (p = 0.002);t = significantly greater than the 10, 75, and 95% oxygen concentrations (p = 0.0001).

Full size image

As shown in Figure 5, the incorporation of 3H label into the explant DNA followed a pattern similar to that observed for bud and branch formation, n = 4 for each oxygen concentration. Lung growth (measured as [3H]thymidine incorporation) in normoxia was significantly greater than any of the other oxygen concentrations studied. There was significantly more growth at 35% oxygen than the other exposures excepting 21%.

Figure 5
figure 5

Incorporation of [methyl-3H]thymidine, 0.5 mCi/mL of DMEM into explants after 48 h of growth at 10-95% oxygen. * = significantly greater than all other exposures, ** = significantly greater than 10, 50, 75, and 95% oxygen.

Full size image

Because of the small amount of growth that occurred with 50% oxygen exposures, that concentration was chosen as the one where the effect of various antioxidants was studied. Of the tested antioxidants, a significant reduction in the growth effects of hyperoxia was achieved by those which detoxified superoxide (Fig. 6,A andB).Figure 6A shows that the addition to the medium of MnSOD(500 U/mL), 25 mM TMPYP, and the combination of CuZnSOD (500 U/mL) and catalase (400 U/mL) resulted in significantly more branch formation compared with explants incubated in 50% oxygen without the added antioxidants. Branch formation with added deferoximine approached but did not reach statistical significance, p = 0.08. There was no significant difference in branch formation between incubations at 21% oxygen, CuZnSOD + catalase, and TMPYP. As can be seen in Figure 6B, the increase in bud formation was also significantly greater for deferoximine, MnSOD, TMPYP, and the combination of CuZnSOD and catalase than in the nonsupplemented 50% exposures. Unlike the branch results, there were significantly more buds formed in 21% oxygen than CuZnSOD + catalase (p = 0.016) and deferoximine (p = 0.041). The difference between 21% oxygen exposures and TMPYP approached, but did not reach statistical significance(p = 0.066). TMPYP in concentrations greater than 100 mM was toxic to the explants. Interestingly, there was no measurable protection at 50% oxygen exposures by either CuZnSOD or catalase alone providing evidence for synergy when they were used together.

Figure 6
figure 6

Branch (A) and bud (B) increase in explants incubated for 48 h in either 21% oxygen, 50% oxygen, or 50% oxygen in the presence of deferoximine, MnSOD, TMPYP, catalase, CuZnSOD, and the combination of CuZnSOD + catalase. (A) a = significantly less than normoxia (p = 0.0001); b = not significantly greater than 50% oxygen; the following were significantly greater than 50% oxygen; c (p = 0.0001); d (p = 0.004);e (p = 0.001). (B) a = significantly less than normoxia (p = 0.0001); significantly greater than 50% oxygen; b (p = 0.007); c (p = 0.02);d (p = 0.0002); e (p = 0.017).

Full size image

Figure 7,A,B,C,andD, shows that similar results were found for TMPYP and deferoximine when lungs from later gestation 14-d fetuses were incubated for 48 h in normoxia or 50% oxygen. Because of the large number of branches and buds, quantification of growth was not done. However, it is apparent that there was protection by both TMPYP and deferoximine against the growth inhibition effect of 50% oxygen in this later gestation lungs.

Figure 7
figure 7

Appearance of 14-d gestation fetal lungs incubated for 48 h in (A) 50% oxygen, (B) 21% oxygen, (C) 50% oxygen with TMPYP, and (D) 50% oxygen with Desferal.

Full size image

DISCUSSION

The hyperoxia administered to premature infants with respiratory distress syndrome is believed to be partially or wholly responsible for the development of BPD. This study shows that exposures of the fetal lung explants to oxygen as low as 35% have an adverse effect on lung growth and development. The injury at 50% and greater oxygen was irreversible after only 24 h of incubation. We were surprised to find that exposures to 50% oxygen completely inhibited bud and branch formation and growth, because this concentration is considered to be safe in the clinical setting. It is possible that the mouse fetal lung explants are more sensitive to hyperoxia than the lungs of premature infants. We also were surprised to find that the explants did not grow in a hypoxic environment, because that concentration is close toin utero, and birth into normoxia has been suggested as representing an “oxidant shock.” We found that branching morphogenesis in 14-d gestation lung explants was also adversely affected by 50% oxygen exposures. Importantly, explants from these more developed lungs demonstrated the same qualitative response to TMPYP and deferoximine as was observed with the 12-d gestation explants.

In our study the explants exposed to 95% oxygen lost all identifiable morphology. This contrasts with the report by Resnick et al.(14), where mouse fetal lung explants incubated in 95% oxygen were morphologically more intact than ours. Their earlier study added chick embryo growth factor and horse serum to the medium, whereas we did not. We speculate that this difference may be an effect of the hyperoxia on as yet undefined growth factors in the chick embryo growth factor and/or horse serum which are important for lung branching morphogenesis. Germane to this speculation is the report that human endothelial cells growing in the presence of endothelial cell growth factor ceased growing when the oxygen concentration was increased from 21 to 50%(16). The authors attributed this to the inactivation of endothelial cell growth factor by the 50% oxygen.

Our observation that exposures to oxygen concentrations as low as 35% produced a measurable inhibition of explant growth and development is the first to suggest that there may not be a critical threshold for the effect of hyperoxia on the developing lung. A 50% oxygen exposure has been suggested as an injury threshold for pulmonary endothelial cells(17). The concept of a threshold for hyperoxia injury to the lung was also suggested in another report which found that 50% oxygen had no effect on the resolution of injury in lungs damaged by oleic acid(18). It is possible that a threshold for oxygen toxicity could exist for adult, but not neonatal, animal lungs.

The concentration of oxygen in the umbilical artery is approximately 7%(19). Thus, our finding that the explants incubated in 10% oxygen did not develop was also a surprise, because mouse pronuclear embryos grow better in 5% oxygen than in normoxia(20). In that study, normoxia incubations resulted in almost no blastulation. Interestingly, the addition of human recombinant CuZnSOD to medium in which the embryos were incubated resulted in a significant improvement in blastulation both at the low and normoxic incubations. Although that study suggested otherwise, it is possible that the mechanism by which hypoxia prevented fetal lung explant growth is different from that of hyperoxia where superoxide appears to be the causative ROS.

We were surprised that, of a wide variety of antioxidants tried in our system, only the combination of MnSOD, CuZnSOD with catalase, TMPYP, a low molecular weight synthetic SOD, and deferoxamine, an antioxidant which is also an inhibitor of hydroxyl radical formation, were effective in reducing the adverse effects of 50% oxygen. Notably, deferoxamine was less effective than either MnSOD, CuZnSOD with catalase, or TMPYP in preventing the effect of hyperoxia on branch growth. These results provide evidence that superoxide anion is partly, if not wholly, responsible for the hyperoxia-induced effect on explant morphogenesis. Transgenic young mice, in whom lung CuZn activity was increased approximately 2-fold, survived longer when exposed to 99% oxygen than did controls(21). No difference in survival was detected in older mice. In another transgenic mouse experiment, Wispeet al.(22) increased MnSOD activity in lung epithelial cells. The transgenic mice survived significantly longer than the nontransgenic controls. Aerosolized MnSOD has also been reported to decrease pulmonary injury in baboons exposed to 100% oxygen for 96 h(23). Our finding that deferoxamine was also successful in mitigating the effect of 50% oxygen is somewhat more difficult to interpret. In addition to preventing the formation of hydroxyl radical through its iron chelating effect, deferoxamine can also be oxidized by superoxide and other strong oxidants(24). In addition, it mayinhibit nitric oxide synthase activity which is iron transcription-mediated(25). Thus, the effect of deferoxamine on mitigating the effects of hyperoxia on the mouse fetal lung explant may be due to mechanisms other than preventing hydroxyl radical formation.

Although we do not propose that the 12-d gestation fetal lung system described in this study is an appropriate model for BPD, it may qualitatively reflect the effect of hyperoxia on a fetal lung in the later saccular stage. The similarity in response after the addition of the SOD mimetic to 14-d gestation compared with the 12-d gestation explants exposed to 50% oxygen suggests that extrapolation to lungs in a later stage of development is possible. Clearly, the best animal model of human BPD is that of premature baboons exposed to prolonged hyperoxia and assisted ventilation(5). However, because of its expense, it is not reasonable to use the baboon model as a general screening tool for BPD interventions. The mouse fetal lung explant therefore may be useful as an inexpensive method to screen compounds for their potential effectiveness in reducing hyperoxia damage to the fetal lung. Those which are found successful would be logical candidates for further study in the baboon model.

This study has demonstrated that the 12-d gestation mouse fetal lung is very susceptible to hyperoxia damage with an adverse effect on branching morphogenesis observed with oxygen exposures as low as 35%. Little to no growth was seen after exposures to concentrations of 50% oxygen and higher. The addition of MnSOD, the combination of CuZnSOD and catalase, and a low molecular weight SOD mimetic to the explants significantly mitigated the adverse effect of 50% oxygen on branching morphogenesis. These results provide evidence that superoxide anion formation is responsible for most of the growth-inhibitory effects of hyperoxia on the mouse fetal lung explant.