Main

Very premature infants are at high risk for bronchopulmonary dysplasia (BPD) (1). Vitamin A (VA; retinol) deficiency often observed in premature infants is associated with a higher incidence of BPD (2). Randomized controlled trials and a recent systematic review indicate VA supplementation decreases BPD and/or death (3,4). Despite supplementation, 25% of infants remain VA deficient (4). The persistence of biochemical VA deficiency may be due to impaired VA transport as transthyretin, a major VA transport protein, is reduced by inflammation (5).

Retinoids are important for normal lung development and maturation (6,7). Hyperoxia induces lung injury, and may contribute to BPD (1). Animal models have shown that administration of retinoic acid (RA, the principal metabolite of VA) attenuates hyperoxic lung injury (8,9). We have shown that a combination of VA (the nutrient) and RA (the metabolite) improved tissue retinoid stores much more than either VA or RA alone in infant rats (10,11). It is possible that the functional consequence of this increase in tissue retinoid stores is a synergism in the attenuation of hyperoxic lung injury, which may be of therapeutic relevance in premature neonates as the administration of RA may circumvent the effects of a reduction in VA transport.

The objective of this study was to determine whether the combination of VA and RA in the newborn mouse model would increase lung retinyl ester content and reduce early hyperoxic lung injury to a greater extent compared with VA or RA administered alone.

METHODS

Experimental animals and animal care.

The study was approved by the Institutional Animal Care and Use Committee (IACUC) and was consistent with the Public Health Service policy on Humane Care and Use of Laboratory Animals (Office of Laboratory Animal Welfare, 2002).

Neonatal mouse model of hyperoxic lung injury (in vivo).

Newborn C57BL/6 mice along with their dam were exposed to hyperoxia (95% O2) or room air from birth to 4 d of age in a Plexiglas chamber. The O2 concentration (OM-100 oxygen analyzer, Newport Medical Instruments, Newport Beach, CA), humidity, temperature, and barometric pressure (Fisherbrand Digital Barometer, Fisher Scientific, Pittsburgh, PA) within the chamber were continuously monitored.

Experimental design.

Newborn mouse pups when exposed to either air or hyperoxia were given one of four agents: vehicle (canola oil), VA, RA, or vitamin A + retinoic acid (VARA) daily for 4 d, and then killed. At least four litters were used for each of these eight groups. Two litters in each group were used for evaluation of lung injury, and the other two litters were used for evaluation of lung and liver retinyl ester content. Additional litters (Air-vehicle, Air-VARA, Hyperoxia-vehicle, and Hyperoxia-VARA groups) were used either for bronchoalveolar lavage fluid (BALF) collection or for calculation of the wet to dry weight ratio.

Administration of vehicle, VA, RA, or VARA.

VA (all-trans-retinyl palmitate, Sigma Chemical Co.-Aldrich, St. Louis, MO) was diluted directly in canola oil to a concentration of 0.1 mmol/g oil (2× concentrate). For RA (all-trans-RA, Sigma Chemical Co.-Aldrich), 200 μL ethanol was added to 3 mg RA, and 800 μL oil was then added to achieve a concentration of 0.01 mmol/g (2× concentrate). The 2× concentrates of VA and RA were mixed 1:1 (wt:wt) with oil to form the 1× doses for administration, or 1:1 with each other to form the VARA dose (0.05 mmol VA: 0.005 mmol RA per g dose). Retinoids were stored at 4°C in foil-wrapped vials. The doses were administered into the mouse pup's mouth at 2 μL per day daily for 4 d. These doses were based on our previous studies in rat pups (10). The dose selected was 20,000 IU (6 mg retinol)/kg for VA (100 IU for a 5-g mouse pup), based on Humphrey et al. (12) who provided 50,000 IU to 2.5 kg neonates. For RA, the dose was 500 μg/kg with an assumed 80% absorption (3.1 μg for a 5 g pup), on the basis of the work of Massaro and Massaro (13).

Evaluation of lung and liver retinol levels.

Lung and liver retinol and retinyl esters (RE) were measured by HPLC (10).

Evaluation of lung injury.

Lung injury was evaluated in at least six animals per group from two litters. Lungs were fixed in inflation (25 cm H2O pressure) using 10% formalin. Ten random high power (400×) fields from 5 μm H&E stained sections of a midcoronal section from apex to base of both lungs were evaluated, using a scoring system by an observer masked to treatment group (Table 1). Lung sections were stained with Pan Macrophage Marker F4/80 (Abcam, Cambridge, MA) and Myeloperoxidase (Thermo Scientific, Fremont, CA) for evaluation of macrophages and neutrophils, respectively. Primary antibodies were used at 1:100 dilutions for 30 min followed by appropriate secondary antibodies conjugated with AlexaFluor 594 (Invitrogen, Eugene, OR). Slides were then evaluated for macrophage and neutrophil abundance using a scoring system (Table 1) by an observer masked to slide identity. Quantitative morphometric analysis was carried out by light microscopy interfaced with an image analysis system (MetaMorph 6.2v4 software, Universal Imaging Corp., Downingtown, PA).

Table 1 Scoring system for lung injury and macrophage abundance
Full size table

BALF was collected by tracheostomy and cannulation with a 24G cannula, and two instillations of 0.3 mL of sterile PBS. BALF was assayed for total protein using the Bradford Assay (Bio-Rad) and for cytoplasmic lactate dehydrogenase (LDH, a marker of cell injury) (Sigma Chemical Co.-Aldrich).

Quantitative real-time PCR for mRNA was performed as described previously (14,15) on homogenized lungs for cytokines known to be involved in hyperoxic lung injury or with BPD/death: IL-1β (16,17), IL-6 (16–18), TNF-α (17–19), IL-10 (16), Monocyte chemoattractant protein (MCP)-1 (20,21), Macrophage inflammatory protein (MIP)-2α (21,22), and regulated on activation, normal T-cell expressed, and secreted (RANTES) (16). The mouse primer sequences are listed in Table 2. Protein concentrations of these cytokines were evaluated in whole lung homogenates by ELISA (R&D Systems, Minneapolis, MN & SA Biosciences, Frederick, MD), and normalized to total protein measured using the Bradford Assay. To batch process all the samples simultaneously for better validity, only the four main groups (21% O2 Vehicle, 21% O2 VARA, 95% O2 Vehicle, and 95% O2 VARA) were evaluated.

Table 2 Mouse primer sequences for real time quantitative PCR
Full size table

Quantitative real-time PCR was performed for peptide growth factors known to be involved in hyperoxic lung injury: PDGF (23), VEGF (24,25), and TGF-β1 (26). These peptide growth factors were also measured by ELISA (R&D Systems) in whole lung homogenates.

Evaluation of lung development.

Alveolar development was evaluated by mean linear intercept (MLI) (27), radial alveolar counts (RAC) (28), and secondary septal crest density (15) as previously described (14,15).

Statistics.

Data are presented as the mean ± SEM. Lung and liver retinyl ester contents, lung developmental indices, and lung injury scores were analyzed by a three-way ANOVA (Factors: Oxygen, VA, RA), followed by multiple comparison testing by the Holm-Sidak method if significant differences (at p < 0.05) were noted by the ANOVA. When the variance terms were unequal among groups, log10 transformation was performed before statistical testing.

RESULTS

The mouse pups exposed to hyperoxia did not have obvious respiratory distress or impaired survival. Administration of retinoids was well tolerated. No differences in growth were noted with hyperoxia or with retinoid administration at 4 d.

Lung retinyl ester concentration.

Seventy mouse pups were evaluated for lung RE (6–15/group). In air-exposed animals, VA and RA both increased lung RE content to a comparable extent, whereas the combination of VA and RA led to a marked synergistic increase in lung RE content, as the increase by VARA was more than the increase by the same amount of VA or RA alone (Fig. 1A). Hyperoxia-exposed animals given vehicle had lower lung RE content, whereas VA and VARA both led to significant and comparable increases in lung RE content. The synergistic effect of the VARA combination on lung RE content that was seen in air-exposed animals was attenuated in hyperoxia. RA alone did not significantly increase lung RE in hyperoxia (Fig. 1A). The relative proportions of lung REs were similar for all treatment groups, showing a major peak of retinyl palmitate with a shoulder of retinyl oleate, followed by a second peak of retinyl stearate (data not shown).

Figure 1
figure 1

Lung and liver retinyl ester concentrations in neonatal mice exposed to room air or hyperoxia, while being treated with vehicle, VA alone, RA alone, or VARA. Lung and liver RE (nmol/g) were analyzed on day 4. Panel A, lung retinyl ester concentrations. [mean ± SEM; n per group shown at base of bar]. Panel B, liver retinyl ester concentrations. [mean ± SEM of log10 transformed data; n per group shown at base of bar] (*p < 0.05 vs corresponding vehicle, †p < 0.05 vs corresponding VA, ‡p < 0.05 vs corresponding RA, §p < 0.05 vs corresponding air).

Full size image

Liver retinyl ester concentration.

Fifty-three mouse pups were evaluated for liver RE (4–7/group). In both air-exposed animals and hyperoxia-exposed animals, VA and VARA moderately increased liver RE content (Fig. 1B). RA did not change liver RE in air-exposed animals, although a small but statistically significant decrease was noted in the hyperoxic mice (Fig. 1B).

Histology.

Sixty-three mouse pups were evaluated by histology (6–13/group). In air-exposed animals, VA, RA, and VARA did not change lung injury scores (Table 3). Hyperoxia increased lung injury scores in vehicle-exposed mice, although the magnitude of injury was mild (average score per field <2) (Table 3; Fig. 2). In the hyperoxia-exposed animals, administration of VA, RA, and VARA lowered injury scores compared with the vehicle group, and there was a further decrease in injury scores in the VARA group compared with the RA group (Table 3; Fig. 2). There was no significant increase in alveolar macrophages in the hyperoxia-exposed mice (Table 3). Very few neutrophils were observed on the myeloperoxidase staining, indicating a lack of neutrophilic infiltrate at this time point. No differences in wet-to-dry weight ratio or lung volumes were noted among the groups (data not shown).

Table 3 Lung injury and macrophage abundance scores
Full size table
Figure 2
figure 2

Photomicrographs of lungs from neonatal mice exposed to room air or hyperoxia while being treated with A, vehicle; B, VA alone; C, RA alone; or D, VARA. (H&E stain; 400×; calibration bar = 50 μm; Br = Bronchus). Exposure to hyperoxia led to hemorrhage (erythrocytes within alveoli and in alveolar septae) and airway epithelial injury (denuded epithelial cells within Br) that were attenuated with VA, RA, and to a greater extent with VARA.

Full size image

Cytokines.

Hyperoxia increased mRNA of IL-1β, IL-6, MCP-1, MIP-2α, and TNF-α, and decreased that of IL-10 (Fig. 3). VARA supplementation prevented the hyperoxia-induced increases of IL-1β, IL-6, and TNF-α but did not affect the MCP-1 increase. The hyperoxia-induced reduction in IL-10 was prevented by VARA. VARA supplementation at 21% O2 did not change gene expression of any of the evaluated cytokines. RANTES gene expression did not change with either hyperoxia or VARA.

Figure 3
figure 3

Cytokine gene expression and protein levels in lungs of neonatal mice treated with vehicle (veh) or VARA while exposed to room air or hyperoxia. Gene expression and protein levels of (A) IL-1β, (B) IL-6, (C) MCP-1, (D) MIP-2, (E) RANTES, (F) TNF-α, and (G) IL-10 in lung homogenates were analyzed on day 4 by competitive real-time PCR and by ELISA, respectively (Mean ± SEM, n = 6 animals per group; *p < 0.05 vs corresponding vehicle, §p < 0.05 vs corresponding air).

Full size image

Changes in protein concentration of these cytokines in the lung homogenates however did not follow the changes in mRNA, with the exception of MIP-2α, which was increased in the hyperoxia-vehicle group and was similar to air vehicle and air VARA in the hyperoxia-VARA group (Fig. 3). IL-1β and RANTES protein were not significantly changed with either hyperoxia or VARA. IL-6 increased with hyperoxia, and VARA did not change IL-6 significantly in either air or hyperoxia compared with vehicle at the same oxygen concentration. Hyperoxia did not alter MCP-1 and TNF-α, and VARA reduced MCP-1 and TNF-α in both air and hyperoxia conditions compared with vehicle. IL-10 protein was increased in the hyperoxia-vehicle group, and this increase was prevented by VARA.

Peptide growth factors.

Neither hyperoxia nor VARA supplementation significantly changed PDGF mRNA or protein in lung homogenates (Fig. 4). VEGF mRNA was increased only in the hyperoxia-VARA group, but VEGF protein was increased in both hyperoxia-vehicle groups and hyperoxia-VARA groups (Fig. 4). A trend toward increased TGF-β1 mRNA expression was noted in hyperoxia-vehicle and hyperoxia-VARA groups (0.05 < p < 0.20 for effect of 95% O2), but total (latent + active) TGF-β1 protein was similar in all groups, and active TGF-β1 protein was reduced in the hyperoxia-vehicle and hyperoxia-VARA groups (Fig. 4).

Figure 4
figure 4

Peptide growth factor gene expression and protein concentration in lungs of neonatal mice treated with vehicle (veh) or VARA when exposed to room air or hyperoxia. Gene expression and protein concentrations of (A) PDGF, (B) VEGF, and (C) TGF-β1 in lung homogenates were analyzed on day 4 by competitive real-time PCR and by ELISA, respectively. (Mean ± SEM, n = 6 animals per group; *p < 0.05 vs corresponding vehicle, §p < 0.05 vs corresponding air).

Full size image

BALF analysis.

LDH was increased in the BALF of the hyperoxia-vehicle group (A490–670: 0.027 ± 0.006 [mean + SEM], p < 0.05 versus other groups), while LDH concentrations were similar in the air-vehicle (0.016 ± 0.002), air-VARA (0.013 ± 0.002), and hyperoxia-VARA (0.013 ± 0.002) groups. Protein concentrations were not significantly different (mean + SEM in μg/mL, air-vehicle: 6.4 ± 1; air-VARA: 7.2 ± 1; hyperoxia-vehicle: 11.8 ± 4; hyperoxia-VARA: 7.2 ± 4; p = 0.2 by 2-way ANOVA).

Lung development.

Changes in MLI and secondary crest density over this brief time period were not statistically significant (p > 0.05) (Table 4). RAC was lower in pups exposed to hyperoxia, but administration of VA, RA, and VARA did not change RAC significantly (Table 4).

Table 4 Lung development assessment
Full size table

DISCUSSION

We observed that an equimolar combination of VA and RA increased lung retinyl ester concentration in a newborn mouse model more than either VA or RA alone, confirming our previous studies in neonatal rats that the VARA combination synergistically increases lung RE content (10). A novel observation was that the synergistic effect of VARA on lung RE content seen in air-exposed newborn mice was blunted in hyperoxia-exposed mice, suggesting increased consumption or decreased uptake of lung RE during hyperoxia. In addition, VARA demonstrated synergism in attenuating early hyperoxic lung injury, accompanied by prevention of hyperoxia-induced increases in MIP-2 mRNA and protein. VARA also prevented hyperoxia-induced increases in gene expression of several pro-inflammatory cytokines (IL-1β, IL-6, and TNF-α) and hyperoxia-induced reduction of IL-10, with less consistent changes in protein concentrations of these cytokines.

Lung development in the human from week 24 of gestation through the first 2 y of age parallels lung development of mice in the first two postnatal weeks (29,30), and the newborn mouse lung is therefore a good model for lung injury in preterm infants. Exposure of neonatal mice to hyperoxia during this period of rapid lung development leads to pathophysiology similar to human BPD with decreased alveolarization (31). Retinoids are essential for normal lung morphogenesis and postnatal maturation (32). Veness-Meehan et al. (9) demonstrated that RA treatment of newborn rats during hyperoxia improves survival and alveolarization. The exact mechanisms by which retinoids attenuate hyperoxia effects in the newborn lung are not certain, but modulation of cytochrome P4501A enzymes (8), growth factor expression (33,34), or CRABP-I (35) may play a role. Regardless of the mechanism by which retinoids exert their effect, an increase in lung RE content is essential for this benefit. Neonates have very low VA reserves due to the limited transplacental VA passage (36), and lung RE content declines abruptly before birth (37). As RA administration increases alveolar septation even in normally nourished rodents (13), physiologically low levels of RE at birth may be marginal and rate limiting for lung development. We have previously shown that the VARA combination is several-fold more effective in increasing lung RE content than either VA or RA separately (10,11). The present study determined that this increase in lung RE content may be of functional significance.

The hyperoxia-induced increases in pro-inflammatory cytokines and reduction in the anti-inflammatory IL-10 mRNA, and the attenuation of these changes with VARA indicates a possible mechanism of action of retinoids. However, except for MIP-2α, we observed that changes in protein concentrations of these cytokines did not correlate with mRNA changes. It is possible that post-transcriptional changes in protein synthesis or stability maybe involved and possibly regulated by retinoids. An alternative explanation is that techniques for measuring mRNA and protein differ in sensitivity. We have observed similar discordance between mRNA and protein estimations of various cytokines at later time points (7 d and 14 d) in the hyperoxic newborn mouse model (unpublished data). These results emphasize the importance of measurement of not just gene expression but also corresponding protein concentrations, as well as the complexity of inflammatory regulation in this model. MIP-2α, the murine equivalent of IL-8 in humans, is a potent neutrophil and macrophage chemoattractant, and blockade of its receptor CXCR2 reduces hyperoxic lung injury in adult animals (38). It is possible that retinoid-induced attenuation of hyperoxic increases in MIP-2 mRNA and protein contribute to its mechanism of action by a reduction in subsequent lung inflammatory cell infiltrate.

We noted that hyperoxia increased VEGF protein and decreased active TGF-β1 in lung homogenates. Increased VEGF has been noted by other investigators from days 4 to 9 followed by a decrease from days 12 to 14 (25), indicating that temporal profiles of these mediators may be important. VEGF protein levels are also initially high followed by a decline in premature infants who subsequently develop BPD/death (24). Studies involving hyperoxia-exposed newborn animal models have demonstrated increased TGF-β signaling (26,39), but these studies have not evaluated early time points. We speculate that differences in TGF-β signaling may be isoform specific and time dependent in the newborn lung, and the initial reduction in active TGF-β may possibly initiate impaired alveolarization, whereas later increased TGF-β signaling may contribute to a profibrogenic state.

The demonstration of VARA synergism is of clinical relevance. Increased VA dosing is not sufficient to alleviate VA deficiency in extremely low birth weight infants with lung inflammation, due to a lack of transport of VA to the lung (5). This study overcomes this hurdle by coadministration of VA and RA. As RA increased lung RE, and RA is not converted to RE, increased lung RE content must be due to regulation of retinol metabolism or distribution by RA. Although RA regulates retinol metabolism and increases RE content modestly, VA is rate limiting for the storage of larger amounts of RE in the lungs (10). In the VARA preparation, the molar ratio of RA to VA is 1 to 10, suggesting that a small relative amount of RA influences the distribution of a larger amount of VA. The synergistic effect of the VARA combination on lung RE content indicates that VA and RA have different but interacting roles in lung RE storage (10). Similar to our previous results in newborn rats, liver RE content increased only modestly and only to the VA component of VARA, possibly due to rapid oxidation of RA in the liver (40). The synergistic effect of VARA on lung RE was blunted with hyperoxic exposure, possibly due to increased RE consumption or decreased uptake, although this hypothesis was not specifically investigated in this study.

However, there are limitations to this study. First, although hyperoxia-exposed newborn mice are a common reproducible animal model, this model may not simulate all aspects of human BPD. Effects of VA and RA in newborn mice may not be very similar to their effects in preterm infants. As our study was designed to evaluate the effects of short-term hyperoxia on lung retinyl esters and early lung injury, we did not observe effects on lung development and noted only mild lung injury and few inflammatory cells. Studies in newborn rats have shown that lung myeloperoxidase is not increased until day 6 of hyperoxia (22), and newborn mice exposed to 80% oxygen for 8 d did not have neutrophil infiltration (41). The rationale for evaluation at the early 4-day time point, even though lung injury is not severe, was to identify the early hyperoxia-induced changes that sets into motion subsequent inflammation/injury. Once lung inflammation is established, it is difficult to determine the sequence and relevance of mediators as many cytokines and growth factors are released by inflammatory cells. Despite these limitations, this study has many strengths, including the analysis of not only biochemical evidence of RE storage but also the functional impact on lung injury in a clinically relevant model.

In conclusion, this study demonstrates the efficacy of a combination of VA and RA in increasing not only lung RE concentrations but also in reducing hyperoxic lung injury in a newborn mouse model. The combination of VA and RA has the therapeutic potential of reducing BPD to a greater extent than with VA supplementation alone. Further investigation is required to determine whether the synergism of VARA combination attenuates other models of lung injury and whether this synergism is limited to the neonatal period.