Impact of Dietary Tomato Juice on Changes in Pulmonary Oxidative Stress, Inflammation and Structure Induced by Neonatal Hyperoxia in Mice (Mus musculus)

Many preterm infants require hyperoxic gas for survival, although it can contribute to lung injury. Experimentally, neonatal hyperoxia leads to persistent alterations in lung structure and increases leukocytes in bronchoalveolar lavage fluid (BALF). These effects of hyperoxia on the lungs are considered to be caused, at least in part, by increased oxidative stress. Our objective was to determine if dietary supplementation with a known source of antioxidants (tomato juice, TJ) could protect the developing lung from injury caused by breathing hyperoxic gas. Neonatal mice (C57BL6/J) breathed either 65% O2 (hyperoxia) or room air from birth until postnatal day 7 (P7d); some underwent necropsy at P7d and others were raised in room air until adulthood (P56d). In subsets of both groups, drinking water was replaced with TJ (diluted 50:50 in water) from late gestation to necropsy. At P7d and P56d, we analyzed total antioxidant capacity (TAC), markers of oxidative stress (nitrotyrosine and heme oxygenase-1 expression), inflammation (interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α) expression), collagen (COL) and smooth muscle in the lungs; we also assessed lung structure. We quantified macrophages in lung tissue (at P7d) and leukocytes in BALF (at P56d). At P7d, TJ increased pulmonary TAC and COL1α1 expression and attenuated the hyperoxia-induced increase in nitrotyrosine and macrophage influx; however, changes in lung structure were not affected. At P56d, TJ increased TAC, decreased oxidative stress and reversed the hyperoxia-induced increase in bronchiolar smooth muscle. Additionally, TJ alone decreased IL-1β expression, but following hyperoxia TJ increased TNF-α expression and did not alter the hyperoxia-induced increase in leukocyte number. We conclude that TJ supplementation during and after neonatal exposure to hyperoxia protects the lung from some but not all aspects of hyperoxia-induced injury, but may also have adverse side-effects. The effects of TJ are likely due to elevation of circulating antioxidant concentrations.


Introduction
Due to improvements in neonatal clinical care, most preterm infants born after 25 weeks of gestation are now capable of survival. However, many preterm infants, especially those born before 32 weeks of gestation, are at high risk of developing bronchopulmonary dysplasia (BPD) [1,2]. BPD is a form of lung injury involving increased inflammation [3,4] and is characterized by alterations in the structure and function of both the gas-exchanging tissue of the lung and the conducting airways [5,6]. It is now recognized that the lung injury and altered lung development associated with BPD increases the risk of persistent respiratory morbidities such as impaired lung function [7,8] and asthma [9,10], as well as reduced exercise capacity [11,12].
In spite of advances in the respiratory management of very preterm infants, the incidence of BPD has not significantly decreased [2]. Approximately 90% of very-and extremely-preterm infants require some form of supplemental oxygen therapy [13] and it is now established that prolonged exposure to supplemental oxygen is a major contributor to BPD [14]. In support of the clinical findings, numerous experimental studies have demonstrated that prolonged inhalation of hyperoxic gas at a time when the lung is still developing can replicate hallmark features of BPD, namely alveolar simplification, increased airway smooth muscle, and inflammation within the lung [15][16][17][18][19][20][21].
An important factor contributing to lung injury and altered lung development induced by neonatal hyperoxia is thought to be an increase in oxidative stress [22]. Preterm infants are likely to be particularly vulnerable to oxidative stress as they have an immature endogenous antioxidant system and are therefore unable to mount an adequate antioxidant response to a hyperoxic environment [23]. Thus, it has been proposed that antioxidant supplementation may protect preterm infants from the injurious effects of hyperoxia by restoring the redox balance [24,25]. Attempts have been made to improve antioxidant defense in preterm infants by dietary supplementation with known antioxidants [26]. However, clinical trials and laboratory experiments using single antioxidants have shown them to be largely ineffective in ameliorating hyperoxia-induced lung injury [24,[27][28][29]. Furthermore, an imbalance between different antioxidants can result in tissue damage [30], suggesting that supplementation with single antioxidants could be harmful. As antioxidant defense involves multiple pathways [30], supplementation with any single antioxidant is unlikely to provide adequate protection against injury caused by oxidative stress. Therefore, it has been proposed that a mixture of antioxidants may be more effective in preventing hyperoxia-induced tissue injury [30]. In this regard, tomato juice (TJ) may be a suitable candidate for dietary antioxidant supplementation as it contains a complex array of antioxidants including anthocyanin, β-carotene, vitamins A, C and E and lycopene [31]. These antioxidants have the ability to scavenge reactive oxygen species (ROS) [32]; lycopene in particular is a potent scavenger due to the large number of double bonds within its structure [33]. Dietary supplementation with TJ has been shown to improve lung function and reduce neutrophil influx in asthmatic adults [34]; it has also been shown, in neonatal rats, to prevent nicotine-induced lung injury [35], which is predominantly mediated by an increase in ROS.
We hypothesized that dietary supplementation with a range of antioxidants contained in commercially available TJ would protect the developing lung from the injurious effects of hyperoxia by increasing antioxidant capacity and decreasing oxidative stress; as oxidative stress and inflammation are intimately linked, and neonatal inhalation of hyperoxic gas has been shown to adversely affect pulmonary immune status [15], we proposed that the effect of hyperoxia on immune status would also be ameliorated. Our objective was to determine whether alterations in oxidative stress, inflammation and lung structure induced by prolonged inhalation of hyperoxic gas during lung development could be prevented by dietary supplementation with TJ, a known source of antioxidants that could safely be administered to preterm infants. Owing to differences in lung development between males and females [36], we also interrogated our data for effects of sex.

Materials and Methods
Our study was approved by the Monash University Animal Ethics Committee (project number: MARP/2011/061) in accordance with the National Health and Medical Research Council of Australia's Code of Practice for the Care and Use of Animals for Scientific Purposes.

Treatment groups
Time-mated pregnant C57BL6/J mice (Mus musculus) were randomly assigned to four treatment groups: 1. a group that breathed room air from birth (Air); 2. a group that breathed 65% O 2 for 7 days after birth (Hyperoxia, Hyp); 3. a group that breathed room air from birth and whose diet was supplemented with TJ (Air+TJ); 4. a group that breathed 65% O 2 for 7 days after birth and whose diet was supplemented with TJ (Hyp+TJ). In the hyperoxia groups, concentrations of O 2 and CO 2 were continuously monitored (Servoflex MiniMP 5200, Servomex, TX) within the housing chambers (floor area 501cm 2 ; Tecniplast, Italy). Each litter was housed individually. The mice were provided with food and water (or TJ) ad libitum. Mice were health checked daily throughout the experimental period. If mice lost weight (~20% reduction in body weight), stopped eating and/or drinking, or displayed reduced physical activity, they were removed from experimental conditions and euthanized via cervical dislocation. No unexpected deaths were encountered. No invasive experimental procedures were performed in this experiment that required the administration of analgesics or anesthetics to ameliorate animal suffering. Within each treatment group, dams and litters were allocated into short-term (7 days) and long-term (56 days) survival studies. Mice are considered to have reached adulthood by 56 days.
Tomato juice supplementation. Starting at 15 days after mating (embryonic day 15; E15d), pregnant mice from the TJ groups were provided with commercially available TJ (Heinz Pty Ltd, Australia). According to the manufacturer, the TJ contained the antioxidants lycopene (59mg/L) and vitamin C (300mg/L). The TJ was diluted in water at a ratio of 50:50; the dams and offspring in the TJ groups had ad libitum access to the diluted TJ in place of drinking water. The diluted TJ was prepared daily and the volumes consumed were recorded daily; the daily volume of water consumed by the Air and Hyp groups was also recorded. For the short-term survival cohort, drinking water was replaced with the diluted TJ from E15d to necropsy at P7d. For the long-term survival cohort, the diluted TJ was provided from day 15 of pregnancy (E15d) to P56d. Before birth, offspring were presumed to have received antioxidants from the TJ via the placenta and after birth they received antioxidants via maternal milk or from drinking TJ.

Plasma and tomato juice carotenoid concentrations
At necropsy at P56d, blood was collected by cardiac puncture and plasma stored at -80°C. Plasma and TJ samples were analyzed to determine concentrations of the carotenoids lycopene, α-carotene, β-carotene, β-cryptoxanthin and lutein/zeaxanthin by high performance liquid chromatography (HPLC), which was conducted in a darkened laboratory under red light at The University of Newcastle, Australia. Plasma samples from offspring in each treatment group were pooled to obtain a sufficient volume for analysis; therefore, only one value per group was obtained. Ethanol:ethyl acetate (1:1) containing the internal standard canthaxanthin was added to each sample, and each sample was then vortexed and centrifuged at 3000xg for 10 minutes at 4°C. The supernatant was removed and collected, and the above process repeated three times, the first two with ethanol:ethyl acetate (1:1) and the third with hexane. The supernatants collected from each centrifugation were pooled and ultra-pure water was added to the mixture, which was then vortexed and centrifuged. The supernatant containing the organic solvents was decanted and any remaining solvents evaporated with nitrogen. The sample pellets were reconstituted in dichloromethane:methanol (1:2 v/v) and chromatography was performed on a Hypersil ODS column (100mm x 2.1m x 5μm) with a flow rate of 0.3mL/min. Carotenoids were analyzed using a mobile phase of acetonitrile:dichloromethane:methanol (containing 0.05% ammonium acetate) (85:10:5 v/v) and a diode array detector (450nm) [37].
Separate animals from each treatment group (n = 6-8 per group) provided samples of lung tissue that were snap-frozen in liquid nitrogen at necropsy and stored at -80°C for subsequent quantitative real-time polymerase chain reaction (qPCR) analyses and determination of total antioxidant capacity (TAC).

Microscopy analyses
Single sagittal sections (5μm) of both the entire left and right paraffin-embedded lungs were used for histological and immunohistochemical analyses. Sagittal sections (5μm) through the entire right paraffin-embedded lung were used for immunofluorescent analysis. For histological and immunohistochemical analyses, stained sections were examined by light microscopy (Nikon Eclipse E400, Nikon, Japan). Color images were captured at a final magnification of x400 using a digital camera (SPOT Insight 4meg Fire Wire Color Mosaic 14.2 Diagnostic Instruments, MI). Images were analyzed using image analysis software (Image-Pro Plus, version 6.2, Media Cybernetics, MD). For immunofluorescent analysis, stained sections were examined by fluorescent microscopy (Zeiss Imager A1, Zeiss, Germany) and images captured with a digital camera (AxioCam MRc5, Zeiss, Germany) linked to image analysis software (AxioVision, version 4.8, Zeiss, Germany). Images were collected at a final magnification of x200. All analyses were conducted in a blinded manner by the same researcher (S.B).

Oxidative stress in the lung parenchyma
Nitrotyrosine staining within lung tissue was used as a marker for oxidative stress [38]. Nitrotyrosine is produced in tissue by tyrosine nitration following the formation of peroxynitrite by nitric oxide and the superoxide free radical [39]. Nitrotyrosine was detected using an established immunohistochemical stain [40]. In each of the left and right lungs, three and five fields of view of lung parenchyma (free of large conducting airways and major blood vessels) were analyzed at P7d and P56d, respectively. The area of lung tissue positively stained for nitrotyrosine (identified by brown staining) was determined and expressed as a percentage of the total area of tissue for each field of view.
Pulmonary heme oxygenase-1 gene expression Heme oxygenase-1 (HO-1) is an inducible catabolic enzyme that is upregulated in response to a state of oxidative stress [41]. Using frozen lung tissue collected at P7d and P56d, we performed total RNA extraction followed by cDNA synthesis and qPCR to determine the gene expression of HO-1 (GenBank accession number NM_010442.2) [40]. At P7d, HO-1 mRNA levels were normalized against the housekeeping gene β-actin (GenBank accession number NM_007393.3) using the comparative delta C T (cycle threshold) method; at P56d, HO-1 mRNA levels were normalized against the housekeeping gene 18S rRNA (GenBank accession number NC_000072.6) [40]. Values were expressed as a fold-change relative to the mean mRNA levels of the Air group at the same age (P7d or P56d).

Total antioxidant capacity of lung tissue
At P7d and P56d, the TAC of small antioxidant molecules including dietary antioxidants and uric acid was measured in samples of frozen lung tissue using a TAC assay kit, according to the manufacturer's instructions (#ab65329, Abcam, UK). The assay measures the ability of a sample to reduce Cu 2+ ions to Cu + . The Cu + ion is chelated with a colorimetric probe producing an absorbance peak at 560nm. Samples were assayed in conjunction with a series of standards, and TAC determined by interpolation of the standard curve.

Macrophages in lung tissue at P7d
We assessed macrophages in the lung as they play a key role in inflammation [42]. Macrophages in the lung parenchyma were identified using an immunofluorescent stain for galectin-3 [40]. In the right lung, three fields of view that were free of major airways and blood vessels were analyzed. Fields of view were randomly selected using a blue filter, which does not distinguish positively stained cells. The number of macrophages (identified by red staining) was expressed as a percentage of the total number of cells in the lung parenchyma for each field of view.

Leukocytes in bronchoalveolar lavage fluid at P56d
We measured the number and type of leukocytes in BALF as they are an indicator of pulmonary innate immunity. BALF was collected at necropsy via cannulation of the trachea and lavaging the airway lumen with 2 x 1mL of saline. Leukocytes within the BALF were isolated, prepared and stained with either 0.4% trypan blue solution or May-Grünwald-Geimsa for the determination of the total number of leukocytes per mL of BALF and differential counts of macrophages, neutrophils, lymphocytes and eosinophils, respectively [40].
Pulmonary interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α) gene expression We measured IL-1β and TNF-α gene expression as an indicator of pulmonary inflammation. Using frozen lung tissue collected at P7d and P56d, we performed total RNA extraction followed by cDNA synthesis and qPCR [40] to determine the gene expression of IL-1β and TNFα (GenBank accession numbers NM_008361.3 and NM_013693.2, respectively; Sigma). At P7d, mRNA levels for the genes of interest were normalized against the housekeeping gene βactin (GenBank accession number NM_007393.3) using the comparative delta C T (cycle threshold) method; at P56d, the mRNA levels were normalized against the housekeeping gene 18S rRNA (GenBank accession number NC_000072.6) [40]. Values were expressed as a foldchange relative to the mean mRNA levels of the Air group at the same age (P7d or P56d).

Lung morphometry
Lung tissue fraction and mean linear intercept (MLI; an indicator of alveolar size) were measured using sections of lung parenchyma stained with Masson's Trichrome. We used 30 fields of view per offspring (15 each from the left and right lungs) that were free of large conducting airways and major blood vessels. The tissue fraction and MLI were determined by superimposing test grids over each field of view [43].
We assessed bronchiolar wall structure as bronchioles are an important determinant of lung function [44] by influencing airflow to and from the alveoli. Bronchioles selected for analysis appeared in near-circular cross-section and had an intact wall that lacked cartilage; selected bronchioles had a basement membrane perimeter (P BM ) ranging from 200-1000μm (approximate diameter: 64-318μm). Bronchiolar wall variables were expressed in relation to the P BM to adjust for differences in bronchiolar size [45]. Wall structure was analyzed for six bronchioles (3 per lung) from each offspring at P7d and ten bronchioles (5 per lung) from each offspring at P56d; the number of bronchioles analyzed was the maximum number that could be consistently obtained at each time-point that met the selection criteria. We used Masson's Trichrome stained sections to determine the area of the bronchiolar epithelium [46] and the number of bronchiolar-alveolar attachments (N BA ); both were expressed in relation to the P BM [46]. Bronchiolar-alveolar attachments were measured because they provide mechanical support to bronchioles [47]. Gordon and Sweet's Reticular Fiber stain was used to determine the area of types I and III collagen fibers (identified by black staining) in the bronchiolar wall, expressed in relation to the P BM [46]. Immunohistochemistry was performed to determine the proportions of proliferating and ciliated cells in the bronchiolar epithelium, expressed in relation to the total number of bronchiolar epithelial cells, using the markers Ki67 and Foxj1, respectively, as previously described [20]. Airway smooth muscle (ASM) in bronchioles was identified immunohistochemically using a primary antibody against α-smooth muscle actin (α-SMA) [40]. The area of α-SMA in the bronchiolar wall (identified by brown staining) was expressed in relation to the P BM .
Pulmonary expression of collagen and α-SMA genes Using frozen lung tissue collected at P7d and P56d, we performed total RNA extraction followed by cDNA synthesis and qPCR [40] to determine the expression of the collagen genes COL1α1, COL1α2 and COL3α1, and the smooth muscle marker α-SMA (GenBank accession numbers NM_007742.4, NM_007743.3, NM_009930.2 and NM_007392.3, respectively; Gene-Works). At P7d and P56d, the mRNA levels for the genes of interest were normalized against the housekeeping genes β-actin (GenBank accession number NM_007393.3) and 18S rRNA (GenBank accession number NC_000072.6), respectively, using the comparative delta C T (cycle threshold) method [40]. Values were expressed as a fold-change relative to the mean mRNA levels of the Air group at the same age (P7d or P56d).

Statistical analysis of data
Numerical data are expressed as mean ± SEM. Statistical outliers were identified using Grubbs' outlier test and removed if detected (alpha = 0.05) [48]. The Shapiro-Wilk test was used to determine if data were normally distributed; if they were not, the data were transformed to produce a normally distributed population. Comparisons between groups were made using a two-way ANOVA, with treatment (p treatment ) and sex (p sex ) as factors. If the ANOVA showed statistical significance, a least significant difference post-hoc test was performed to identify differences between individual treatment groups. Comparison between treatment groups for the volume of TJ or water consumed over time was made using two-way repeated measures ANOVA. TAC and gene expression of IL-1β, TNF-α, COL1α1, COL1α2, COL3α1 and α-SMA were analyzed using a one-way ANOVA using treatment as a factor. Statistical significance was accepted at p<0.05.

Consumption of tomato juice and water
The volumes of either TJ or water consumed by the dams between birth (P0d) and weaning (P21d) were not significantly different between treatment groups. The mean volume of liquids consumed by all mice (dams and offspring) over this period was 436.7 ± 59.5mL/day/kg body weight. Between weaning and P56d, the volume of either TJ or water consumed by the offspring was not significantly different between groups. The mean volume of liquids consumed in all groups over this period was 261.9 ± 24.0mL/day/kg body weight.

Plasma and tomato juice carotenoid concentrations
The lycopene concentration in the plasma of the TJ groups was higher than in the non-TJ groups (Air, 0.0μg/L; Hyp, 0.0μg/L; Air+TJ, 4.6μg/L; Hyp+TJ, 6.1μg/L). The concentration of lycopene in undiluted TJ was 40mg/L; therefore the concentration that was consumed was 20mg/L. Based on the mean volume of liquids consumed by all mice, the dose of TJ was 8.7 ± 1.2mg/day/kg body weight prior to weaning and 5.2 ± 0.5mg/day/kg body weight after weaning. Plasma concentrations of other carotenoids assayed (i.e. α-carotene, β-carotene, β-cryptoxanthin and lutein/zeaxanthin) were 0.0μg/L (i.e. undetectable) in all groups.

Leukocytes and inflammatory markers in the lung
Galectin-3 staining in lung tissue at P7d. The proportion of cells in the lung parenchyma that were macrophages was significantly greater in the Hyp group than in the Air group   1.7 ± 0.1%; Hyp, 2.2 ± 0.2%; Air+TJ, 1.2 ± 0.2%; Hyp+TJ, 1.6 ± 0.2% (Fig 4A-4E). No sex difference was observed.
Leukocytes in bronchoalveolar lavage fluid at P56d. The total number of leukocytes per mL of BALF was significantly greater in the hyperoxia groups than in the normoxia groups (p<0.01). There were 150% more leukocytes in the Hyp group compared to the Air group (p = 0.004) and 109% more in the Hyp+TJ group compared to the Air+TJ group (p = 0.006). Values were: Air, 9,238 ± 1,304 cells/mL; Hyp, 23,119 ± 3,453 cells/mL; Air+TJ, 12,371 ± 2,130 cells/mL; Hyp+TJ, 25,878 ± 4,054 cells/mL (Fig 5A). No sex difference was observed.
Enumeration of the different populations of leukocytes in BALF showed that there were no significant differences between treatment groups for the proportion of macrophages (overall mean, 98.93 ± 0.25%), lymphocytes (overall mean, 0.48 ± 0.09%) or neutrophils (overall mean, 0.59 ± 0.21%) (Fig 5B-5D, respectively). No eosinophils were observed in any of the treatment groups. No sex differences were observed. IL-1β and TNF-α gene expression in lung tissue at P7d and P56d. At P7d, there was a tendency for the mRNA expression of IL-1β to be lower in the Air+TJ group than in the other three treatment groups; however, this did not reach statistical significance (p = 0.081; Air, At P56d, the mRNA expression of IL-1β was significantly lower in the Hyp, Air+TJ and Hyp+TJ groups compared to the Air group (p = 0.021; Air, 1.0 ± 0.2; Hyp, 0.4 ± 0.1; Air+TJ, 0.5 ± 0.1; Hyp+TJ, 0.5 ± 0.1, Fig 5E). The mRNA expression of TNF-α was significantly greater in the Hyp+TJ group compared to the Air, Hyp and Air+TJ groups (p<0.001; Air, 1.0 ± 0.2; Hyp, 0.8 ± 0.1; Air+TJ, 0.8 ± 0.2; Hyp+TJ, 1.6 ± 0.4, Fig 5F).

Lung morphometry
Tissue fraction. At P7d, lung tissue fraction was significantly lower in the hyperoxia groups than in the normoxia groups (p<0.001; Table 1 Fig 6A). There was a tendency for the gene expression of COL1α2 (p = 0.062) and COL3α1 (p = 0.052) to be greater in the TJ groups compared to the Hyp group, but this did not reach statistical significance (Table 1, S3A and S3B Fig). At P56d, the area of collagen in the outer bronchiolar wall in the TJ groups was significantly greater than in the Air group (p<0.05; Table 2, Fig 6B-6F). Collagen area was significantly greater in females than in males (p sex = 0.026; males, 3.5 ± 0.2μm 2 /μm; females, 3.8 ± 0.1μm 2 /μm). The gene expression of COL1α1, COL1α2 and COL3α1 in lung tissue was not significantly different between groups at P56d ( Table 2, S5A-S5C Fig).
Bronchiolar and pulmonary α-SMA. At P7d, the area of α-SMA in the outer bronchiolar wall and the gene expression of α-SMA in lung tissue did not significantly differ between groups (Table 1, S2H and S3C Figs, respectively). At P56d, the Hyp group had significantly more smooth muscle in the outer bronchiolar wall than the other groups (p<0.001; Table 2, Fig 7). No significant sex differences in the area of α-SMA in the outer bronchiolar wall were observed at P7d or P56d. There was a tendency (p = 0.086) for the gene expression of α-SMA to be greater in the Hyp+TJ group than in the Hyp group, but this did not reach statistical significance ( Table 2,

Discussion
Our study shows that ingestion of TJ attenuates some of the adverse effects of neonatal inhalation of hyperoxic gas. At P7d, TJ ameliorated the hyperoxia-induced increases in the oxidative stress marker nitrotyrosine and in the proportion of macrophages in lung tissue. At P56d, TJ consumption led to increased TAC of the lung tissue and ameliorated the hyperoxia-induced increase in ASM.

Antioxidant concentrations
Few experimental studies [34,35,49,50] have examined the protective effects of multiple dietary antioxidants on induced alterations in lung development. We used TJ to protect against hyperoxia-induced lung injury, as it is reportedly a readily available source of multiple antioxidants that could potentially be administered to preterm infants [31]. The TJ we used had a measured lycopene concentration of 40mg/L; therefore, the diluted TJ provided to the animals had a lycopene concentration of 20mg/L, which equated to an approximate dose of 7mg/day/ kg body weight and led to a plasma lycopene concentration of 5-6ng/L at P56d. The calculated dose of lycopene in our study is equivalent to that used in a previous study that showed a significant protective effect of lycopene supplementation on allergic inflammation in a mouse model of allergic airways disease [51]. There is little evidence regarding the plasma antioxidant concentrations required to exert significant effects in vivo. Although other carotenoids were not detected in the plasma of supplemented mice, TJ increased the TAC of small antioxidant molecules in lung tissue; this increase could be due to antioxidants other than lycopene being increased, especially vitamin C as it reportedly has a concentration of 300mg/L in undiluted TJ, which would equate to an approximate dose of 52mg/day/kg body weight. Our TJ dosing regimen was chosen as it could potentially be given to preterm infants and has been shown to protect neonatal rats from nicotine-induced changes in lung architecture, which are largely due to oxidative stress [35]. In future, it may be beneficial to alter the route or timing of TJ administration to further increase the plasma concentrations of antioxidants in TJ, including lycopene and vitamins A, C and E.

Total antioxidant capacity of lung tissue
We conducted a TAC assay to determine whether TJ led to a functional increase in the TAC of small antioxidant molecules within the lung. At P7d, TAC was significantly greater in the Hyp +TJ group than in the Air group, and there was a tendency for TAC to be increased in the Air +TJ group. These results show that maternal TJ ingestion increases pulmonary TAC in suckling offspring, confirming that the ingested antioxidants are transferred across the placenta and/or are present within the breast milk of the dam. At P56d, the TAC of lung tissue was significantly greater in both of the TJ groups compared to the non-TJ groups. Together, the results obtained at P7d and P56d suggest that dietary TJ increased the lung's ability to reduce ROS. Furthermore, this increase in TAC was related to evidence of significantly lower oxidative stress in the lungs, as discussed below.

Oxidative stress in the lung
It is well established that hyperoxia leads to oxidative stress within the lung [49,[52][53][54], and our observations confirm this effect as there was increased oxidative stress in the Hyp group at P7d. Importantly, HO-1 was decreased in the Hyp+TJ group compared to the Hyp group, and the hyperoxia-induced increase in nitrotyrosine was inhibited by TJ. Further, a recent study has shown that dietary supplementation with vitamin A and retinoic acid attenuates hyperoxia-induced oxidative stress in the lung [49]. In contrast, the use of single antioxidants is largely ineffective at reducing hyperoxia-induced oxidative stress within the lung [55,56]. It appears therefore that multiple dietary antioxidants are more effective in reducing ROS, and hence oxidative stress, than single antioxidants. We found that the antioxidative effects of TJ were present at P56d, with TJ resulting in a decrease in nitrotyrosine deposition in the absence of changes in HO-1. HO-1 and nitrotyrosine are differentially regulated, which is likely the reason for the differing time-course of expression. This appears to be the first demonstration that TJ can induce a decrease in oxidative stress that persists to adulthood. However, this long-term effect may be cause for concern as ROS production is important for normal physiological processes, including cell signaling pathways [57] and host defense [58]. Thus, a depletion of ROS and a reduction in oxidative stress under basal conditions could alter the redox balance within the lung, increasing the redox capacity to an undesirable level, which could be deleterious to normal lung development.

Leukocytes and inflammatory markers in the lung
At P7d, we found a greater proportion of macrophages in the lung parenchyma in the Hyp group than in the Air group. This is consistent with other studies showing that neonatal hyperoxia leads to an increase in pulmonary leukocytes, including macrophages, immediately after exposure [15,59]. However, our Hyp+TJ group did not have an increased proportion of macrophages compared to the Air or Air+TJ groups, suggesting that TJ may protect the lung from the hyperoxia-induced influx of macrophages immediately following exposure. Interestingly, in the Air+TJ group, the proportion of macrophages was reduced, suggesting that TJ affects pulmonary immune cell number under normal physiological conditions, which may impair innate immunity. To further characterize the immune response at P7d, we analyzed the gene expression of the pro-inflammatory markers IL-1β and TNF-α in lung tissue. The finding that there was no significant difference in the expression of IL-1β or TNF-α between treatment groups was unexpected as macrophages are a major source of IL-1β and TNF-α in the lung [60,61] and are an essential component of host defense. However, as we did not interrogate macrophage phenotype and function in the present study, the different proportions of macrophages found between treatment groups could be due to cytokines other than IL-1β or TNF-α being differentially expressed.
At P56d, we confirmed that neonatal hyperoxia increased the concentration of leukocytes in BALF [40], and that this effect was not ameliorated by TJ. We also found that the gene expression of IL-1β was lower in the Hyp, Air+TJ and Hyp+TJ groups than in the Air group and the gene expression of TNF-α was greater in the Hyp+TJ group than in the other three treatment groups. As TNF-α is known to play a role in M1 differentiation, the macrophage population in the Hyp+TJ group may be skewed towards a M1 phenotype, indicating a potentially detrimental increase in inflammation [62]. Taken together, it appears that TJ did not attenuate the longterm pulmonary inflammatory response to neonatal hyperoxia and may have increased part of the inflammatory response.

Lung morphometry
Our findings on the effects of hyperoxia on lung tissue fraction, MLI and bronchiolar-alveolar attachments are consistent with previous studies [15,18,38,63]. We confirm that neonatal inhalation of 65% O 2 inhibits alveolarization, and suggest that TJ does not protect the lung from the adverse effects of hyperoxia on alveolarization. Our results show that the early morphometric effects of hyperoxia (65% O 2 ) do not persist to adulthood, apart from a reduction in the number of bronchiolar-alveolar attachments in the Hyp+TJ group. Overall, our findings indicate recovery from the hyperoxia-induced inhibition of alveolarization by adulthood and suggest that TJ may adversely affect the ability of alveolarization to recover from early impairment. As previous studies have observed changes in lung function in adulthood following exposure to neonatal hyperoxia [19], further studies are required to determine whether TJ alters lung function.
Our findings suggest that TJ does not prevent hyperoxia-induced proliferation of bronchiolar epithelial cells at P7d. As the observed increase in bronchiolar epithelial cell proliferation was not accompanied by an increase in epithelial area, it is possible that the rate of epithelial cell apoptosis was increased, leading to increased cell turnover. At P56d, the proportion of ciliated epithelial cells was decreased in both of the TJ groups, suggesting that TJ alone, but not hyperoxia, could be causing this effect. Thus, the TJ supplemented mice may have had a reduced ability to move mucus along their airways.
The gene expression of COL1α1 in lung tissue and bronchiolar collagen deposition were elevated in the TJ groups relative to the Air group, at P7d and P56d respectively. This TJ-induced increase in collagen was unexpected and could be detrimental to lung function as increased collagen is indicative of pulmonary fibrosis [64] and is observed in asthmatics [65].
An increase in bronchiolar smooth muscle was observed in the Hyp group at P56d, relative to the Air and Hyp+TJ groups. This increased bronchiolar smooth muscle was not accompanied by an increase in the gene expression of α-SMA at either P7d or P56d; this finding may have been because α-SMA deposition was only measured in the outer airway wall whereas α-SMA gene expression was measured in samples of whole lung tissue. The observed increase in ASM in adult mice following neonatal hyperoxia is consistent with previous studies [17,19,66,67]. Increased ASM is a hallmark feature of BPD [68] and is also implicated in the pathogenesis of obstructive lung diseases, especially asthma [69]. As our data indicate that TJ may prevent the persistent increase in ASM induced by neonatal hyperoxia, it could be beneficial in the management of very preterm infants who are at risk of developing BPD and asthma later in life [1,70].

Effects of TJ alone on lung development
Our study revealed differences between the Air and Air+TJ groups at P7d and P56d, suggesting that TJ alone may alter lung development. The potentially adverse effects of TJ observed at P56d may be due to the long duration of TJ ingestion. Our findings suggest that TJ alone can alter cellular and structural development of the lungs, including increasing lung tissue fraction and collagen deposition, and markers of oxidative stress and inflammation under physiological conditions. Therefore, it is plausible that a reduction in ROS resulting from TJ ingestion during normal physiological conditions could adversely affect lung development.

Conclusions
We conclude that, although TJ does not ameliorate hyperoxia-induced alterations in lung architecture immediately following exposure, it has the beneficial effects of increasing TAC and attenuating the hyperoxia-induced influx of macrophages and increase in oxidative stress in the lungs. Importantly, TJ also protected hyperoxia-exposed lungs from a persistent increase in bronchiolar smooth muscle. However, TJ may alter the persistent pulmonary inflammatory response to neonatal hyperoxia, and it appears that TJ alone adversely affects some aspects of lung cellular and architectural development, redox balance and innate immunity, and may induce fibrosis. In view of the potentially adverse effects of TJ observed in our study, future studies could trial lower dosages, different dosing regimens or other routes of administration.