Contents lists available at ScienceDirect Free Radical Biology and Medicine journal homepage: www.elsevier.com/locate/freeradbiomed Original article Hyperoxia induces epigenetic changes in newborn mice lungs Miroslaw Bik-Multanowskia,⁎,1, Cecilie Revhaugb,1, Agnieszka Grabowskaa, Artur Dobosza, Anna Madetko-Talowskaa, Magdalena Zasadac, Ola Didrik Saugstadb a Department of Medical Genetics, Faculty of Medicine, Jagiellonian University Medical College, ul. Wielicka 265, 30-663 Krakow, Poland bDepartment of Pediatric Research, University of Oslo and Oslo University Hospital, Norway c Department of Pediatrics, Faculty of Medicine, Jagiellonian University Medical College, Krakow, Poland ARTICLE INFO ABSTRACT Keywords: Hyperoxia Bronchopulmonary dysplasia DNA methylation Alteration of TGF-β pathway Programming factor in mice Supplemental oxygen exposure is a risk factor for the development of bronchopulmonary dysplasia (BPD). Reactive oxygen species may damage lung tissue, but hyperoxia also has the potential to alter genome activity via changes in DNA methylation. Understanding the epigenetic potential of hyperoxia would enable further improvement of the therapeutic strategies for BPD. Here we aimed to identify hyperoxia-related alterations in DNA methylation, which could affect the activity of crucial genetic pathways involved in the development of hyperoxic lung injury. Newborn mice (n=24)wererandomized to hyperoxia (85% O2) or normoxia groups for 14 days, followed by normoxia for the subsequent 14 days. The mice were sacrificed on day 28, and lung tissue was analyzed using microarrays developed for the assessment of genome methylation and expression profiles. The meanDNAmethylation level was higher in the hyperoxia group than the normoxia group. The analysis of specific DNA fragments revealed hypermethylation of> 1000 gene promoters in the hyperoxia group, con firming the presence of the DNA-hypermethylation effect of hyperoxia. Further analysis showed significant enrichment of the TGF-β signaling pathway (p=0.0013). The hy permethylated genes included Tgfbr1, Crebbp, and Creb1, which play central roles in the TGF-β signaling pathway and cell cycle regulation. Genome expression analysis revealed in the hyperoxia group complementary downregulation of genes that are crucial for cell cycle regulation (Crebbp, Smad2, and Smad3). These results suggest the involvement of the methylation of TGF-β pathway genes in lung tissue reaction to hyperoxia. The data also suggest that hyperoxia may be a programming factor in newborn mice.

1. Introduction Supplemental oxygen administered to premature babies might in jure the lungs [1] and, consequently, contribute to the development of bronchopulmonary dysplasia (BPD). However, the pathogenesis of BPD, which is a serious complication of prematurity, is complex and includes a range of external risk factors, as well as genetic susceptibility. Func tional alterations of genomic pathways related to several genes en coding important growth factors, such as vascular endothelial growth factor (VEGF), transforming growth factor-β (TGF-β), and insulin-like growth factor (IGF), are among key factors responsible for such sus ceptibility [2]. VEGF regulates endothelial cell differentiation and angiogenesis and plays a central role in the formation of embryonic vasculature [3]. TGF-β is involved in inhibition of branching morpho genesis and alveolarization in embryonic lung development [4]. IGF is also involved in growth and injury repair processes in many organs, including the lungs [5]. It is well known that oxidative stress due to reactive oxygen species (ROS) and reactive nitrogen species (RNS) can (chemically) damage lung tissue. However, hyperoxia also has the potential to alter the genome activity in lung cells by inducing DNA modifications. ROS/RNS modify cytosine with the oxidative conversion of 5-methylcytosine (5 mC) to 5-hydroxymethylcytosine. Peroxides can also modify cytosine to 5-chlorocytosine, which mimics 5-mC [6,7]. The above changes induce Abbreviations: BPD, bronchopulmonary dysplasia; VEGF, vascular endothelial growth factor; TGF-β, transforming growth factor-β; IGF, insulin-like growth factor; 5-mC, 5-methyl cytosine; ROS, reactive oxygen species; RNS, reactive nitrogen species; DAVID, Database for Annotation, Visualization and Integrated Discovery; KEGG, Kyoto Encyclopedia of Genes and Genomes ⁎ Corresponding author. E-mail address: miroslaw.bik-multanowski@uj.edu.pl (M. Bik-Multanowski). 1 The authors contributed equally to this study. https://doi.org/10.1016/j.freeradbiomed.2018.04.566 Received 8 February 2018; Received in revised form 20 April 2018; Accepted 21 April 2018 M. Bik-Multanowski et al. improper DNA methylation by inhibiting the binding of DNA methyl transferase 1 (Dnmt1) to DNA [8,9]. The subsequent alteration of the methylation pattern within the CpG sequences can, in turn, result in gene silencing [10]. The above processes might serve as an example of epigenetic regulation of genome activity. Understanding the epigenetic potential of hyperoxic lung injury might enable further improvement of therapeutic strategies. Therefore, in this study, we aimed at identifying specific, hyperoxia-related al terations of DNA methylation, which could affect the activity of crucial genetic pathways involved in the development of hyperoxic newborn lung injury. To achieve this, we used an established model with new born mice exposed to long-term hyperoxia [11] and performed whole genome methylation analysis with complementary expression assess ment of specific genes of interest in lung tissue.
2. Material and methods 2.1. Animal experiment A total of 24 newborn mice (C57Bl/6Tac) were randomized to hy peroxia (85% O2; 12 animals) or normoxia (21% O2; 12 animals) groups for 14 days, followed by normoxia conditions for all animals for the subsequent 14 days. All mice had free access to food and water and were kept under standard conditions in A-Chambers (O2– monitor ProOX110, CO2– monitor ProCO2 P120, BioSpherix). Lung tissue was harvested on day 28 after euthanasia with a zolazepam/tiletamine/ xylazine/fentanyl cocktail. Tissue samples were snap-frozen in liquid nitrogen immediately after cessation of circulation for the subsequent analysis. The experiments were carried out in Oslo and approved by the Norwegian board of animal research welfare (NARA 50/13-5458). 2.2. Microarray methylation analysis Genomic DNA was isolated from tissue samples using a MasterPure DNA Purification Kit (Epicentre). DNA concentration and purity were measured using a UV spectrophotometer NanoDrop 1000 (Thermo Scientific). A total of 5µg of purified genomic DNA was resuspended in 250µl of PBS and sonicated using an ultrasonic processor UP100H (Hielscher) at 100% amplitude for 5min, in an ice-water bath. The sonication conditions were optimized to obtain DNA fragments ranging between 200 and 1000 base pairs in size. The degree of DNA frag mentation was determined using an Agilent 4200 TapeStation Instrument. After sonication and fragmentation analysis, the volume of each sample was adjusted to 250µl with phosphate-buffered saline. Next, 200µl was used for immunoprecipitation of methylated frag ments and the remaining 50µl as a genomic DNA reference. The subsequent microarray analysis of the genome methylation pattern was performed using Mouse CpG Island microarrays, enabling assessment of 88,737 probes representing specific genome methylation sites in, or within 95 base pairs from, 15,342 CpG islands. The micro array experiment was performed according to Agilent Microarray Analysis of Methylated DNA Immunoprecipitation protocol (version 2.3.1). Subsequently, we used the Genomic Workbench software (Agilent) for analysis of methylation data. The above software calculates nor malized Z-score from the Gaussian distribution to effectively judge the methylation status of a given probe on the array. The combined Z-score, which is a summation of the left and right Gaussian Z-scores, reflects the location of a probe log-ratio value in relation to the Gaussian dis tribution of probes with similar temperature melting. A strong positive value of the combined Z-score means that a given probe is methylated, and a strong negative value means that it is unmethylated. Next, we calculated the mean combined Z-scores for each probe and the average for all mean combined Z-scores for the hyperoxia and normoxia groups. Then, we assessed the differences between groups for each genomic fragment. To decrease the probability of false-positive findings, we focused only on those probes for which the calculated difference exceeded three standard deviations. Eventually, we obtained a final list of probes-of-interest. Subsequently, we assessed the list using the Database for Annotation, Visualization and Integrated Discovery (DAVID) [12] to identify differentially methylated genes and genomic pathways. 2.3. Complementary expression analysis of genes of interest Total RNA was extracted from the remaining fragments of lung tissue using an RNeasy Mini Kit (QIAGEN). Subsequently, RNA samples were used for expression analysis of previously identified genomic pathways, which revealed significant methylation differences between hyperoxia and normoxia groups. We applied SurePrint G3 Mouse Gene Expression 8×60K microarrays (Agilent Technologies), according to the manufacturer's protocol. The expression profiles of the genes of interest were compared between the normoxia and hyperoxia groups using Partek computer software for microarray data analysis (www. partek.com). Statistically significant (p < 0.05) expression differences exceeding 20% (Fold Change>1.2) were further evaluated. Fig. 1. The observed hyperoxia-related lung damage. Mouse lung tissue at 100× magnification: normal development (normoxia) on the left; arrested alveolarization (hyperoxia) on the right. M. Bik-Multanowski et al.
3. Results Histological assessment of lung tissue samples confirmed the pre sence of oxygen-related lung damage in the hyperoxia group. Fig. 1 shows arrested alveolarization with enlarged alveolar spaces and sim plified alveolar structure resulting from prolonged hyperoxia. The mean methylation level of all microarray probes was increased in the hyperoxia group compared to the normoxia group (average va lues for mean combined Z-scores 0.57 and −0.31, respectively). This suggests the presence of an overall DNA-hypermethylation effect of hyperoxia, with a resulting shift of the methylation status of the ma jority of gene promoters from “unmethylated” in the normoxia group towards “methylated” in the hyperoxia group. The analysis of specific DNA fragments revealed 1259 probes that were hypermethylated in the hyperoxia group and 252 probes that were hypermethylated in the normoxia group (Supplementary data). The linearity of the Q-Q (quantile-quantile) plots of the mean combined Z scores in each group suggests the normal distribution of the data (Fig. 2). Further analysis using DAVID software identified 1011 mouse genes among 1259 probes with increased methylation in the hyperoxia group. Genomic pathway analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway database [13] revealed statistically sig nificant enrichment of five genomic KEGG pathways (Pathways in cancer, TGF-β signaling pathway, Hippo signaling pathway, Melano genesis, and Prostate cancer pathway) with the greatest gene enrich ment ratio for the TGF-β signaling pathway (p=0.0013 with False Discovery Rate correction for multiple comparisons; 18 out of 85 genes in the pathway identified; Fold Enrichment ratio = 4.1). The hypermethylated genes included the Tgfbr1 gene, which is pi votal for the pathway, as well as Crebbp and Creb1 genes, which play central roles in cell cycle regulation. The hypermethylated elements of Fig. 2. Q-Q (quantile-quantile) plots of combined Z-scores in the hyperoxia and normoxia groups, suggesting the normal distribution of data. Fig. 3. Methylation pattern in the hyperoxia group; TGF-β pathway scheme is taken from KEGG pathway database. theTGF-βsignalingpathwayareshowninFig.3. Also,atotalof227(mouse)geneswereidentifiedamongthe252 probeswithincreasedmethylationinthenormoxiagroup.However,no significantmethylationdifferenceswereobservedforspecificgenomic pathways. Microarray methylation data have been deposited in the ArrayExpress database at EMBL-EBI (www.ebi.ac.uk/arrayexpress) underaccessionnumberE-MTAB-6489. The complementaryexpressionanalysis of TGF-βpathwaygenes revealed significant downregulation of five genes in the hyperoxia group.Threeofthesegenes(Crebbp,Smad2,andSmad3)arecrucialfor cellcycleregulation.Higherexpressionof theabovegenesinthenor moxiagroup functionallycorrespondswith thepreviouslydescribed hypermethylationof genes in thehyperoxiagroup. The geneswith differentmethylationpatterns anddifferential expression levels are listedinTable1andinFig.4. In addition, we observedhighpercentage of genes involved in apoptosisregulationamongallgenesexhibitingstatisticallysignificant expressiondifferencesbetweenhyperoxiaandnormoxiagroups(45out of 470genes). Themajorityof themwereapoptosis inhibitorswith decreasedactivityinthehyperoxiagroup.Wealsoidentified10genes involvedinreactiontooxidativestress.Thelistsoftheabovegenesare availableintheSupplementaryTable. 4.Discussion Hyperoxia-relatedalterationsofcell signalingandgenomeexpres sionhavebeen thoroughlydescribedandreviewed in the literature [14–17]. Itwasshownthat theimmaturityof theantioxidantdefense system, especially in extremely preterminfants [18,19],may be a predisposingfactor toBPD,as theuseofhighoxygenconcentrations, andpositivepressureventilationuponresuscitationandintheneonatal intensivecareunit.However, lessisknownabout theinfluenceofhy peroxia on DNAmethylation. In a recent study, widespread hy permethylationofDNAwasdemonstratedinaratmodelofBPD[20]. OtherstudieshavereportedincreasedglobalDNAmethylationlevelsin humancellsincaseofshort-termtreatmentwithoxygen,whereaslong termtreatmenthadtheoppositeeffect[21].However,epigeneticme chanismsarecomplex, andtheDNAmethylationlevelmightdepend notonlyonexternal factors, suchas theconcentrationofoxygenand durationofhyperoxia,butalsofromtissue-specificresponsestooxygen. An important obstacle instudyingmethylation-relatedepigenetic mechanismsusingnext-generationsequencingisthelackofestablished algorithms for identificationof functionally important alterations of DNAmethylation among verymany hypermethylatedDNA regions [21]. Therefore, inour study,we focusedonassessment of theme thylationofgenepromoters,whichisaclassicalregulatorymechanism ofgeneexpression[22].Weappliedmicroarraysdevelopedspecifically formethylationstudiesandvalidatedourfindingswithcomplementary geneexpressionanalysis. Thepresent study is, toour knowledge, thefirst toassess lung specificDNAmethylationandgenomeexpressioninthecontextofBPD pathogenesis. Basedon these data, we suggest that hyperoxia is a programming factor in the newborn lungwith potentially lifelong consequences.Oneweaknessofthisstudyisthatwehavenotseparated thedifferentcellsinthelungand,consequently,donotknowwhichcell typesareaffected. Lung injuryandrepairprocesses involvemanycellularactivities, includingcellgrowth,differentiation,andremodelingofextracellular matrixcomponents.TGF-βbelongstoagroupofpivotalgrowthfactors Table1 TGF-betapathway-relatedgenesthatexhibiteddifferencesbetweengroupsinmethylationandexpression. Methylationanalysis–hypermethylatedTGF-betapathwaygenesinhyperoxiagroup Genesymbol Genename Genomiclocationof themethylatedDNAfragment (accordingtoNCBI37/mm9) MeanZ-scorein hyperoxiagroup MeanZ-scorein normoxiagroup MeanZ-score difference(SD) Bambi BMPandActivinMembrane BoundInhibitor chr18:3508199–3508243 3.79 0.63 3.16 Bmp4 BoneMorphogeneticProtein4 chr14:47007792–47007836 9.73 6.56 3.17 chr14:47007582–47007626 6.17 2.74 3.43 Bmp6 BoneMorphogeneticProtein6 chr13:38437991–38438035 9.67 6.52 3.15 Bmpr2 BoneMorphogeneticProtein ReceptorType2 chr1:59820636–59820681 6.82 2.79 4.03 Crebbp CREBBindingProtein chr16:4212906–4212950 6.57 2.97 3.6 chr16:4213107–4213151 6.97 3.33 3.64 E2f5 E2FTranscriptionFactor5 chr3:14611165–14611209 5.5 2.11 3.39 Ep300 E1ABindingProteinP300 chr15:81416218–81416262 7.06 3.9 3.16 Gdf7 GrowthDifferentiationFactor 7 chr12:8304878–8304922 4.99 1.94 3.05 chr12:8305295–8305339 7.79 3.99 3.8 Id1 InhibitorofDNABinding1 chr2:152560730–152560774 5.72 2.46 3.26 Id3 InhibitorofDNABinding3 chr4:135700657–135700701 5.06 2.05 3.01 Id4 InhibitorofDNABinding4 chr13:48356696–48356740 5.08 2.08 3.0 Mapk1 MitogenActivatedProtein Kinase1 chr16:16983547–16983591 7.16 3.07 4.09 Pitx2 PairedLikeHomeodomain2 chr3:128903185–128903231 2.78 −0.41 3.19 Ppp2cb ProteinPhosphatase2 CatalyticSubunitBeta chr8:34710437–34710481 5.25 1.73 3.52 Rhoa RasHomologFamilyMember A chr9:108208556–108208600 4.94 1.82 3.12 Smad7 SmadFamilyMember7 chr18:75528547–75528591 5.64 2.02 3.62 Tgfbr1 TransformingGrowthFactor BetaReceptor1 chr4:47365606–47365650 6.06 2.63 3.43 Tgif2 TGFBInducedFactor Homeobox2 chr2:156666261–156666305 5.6 2.5 3.1 Complementaryexpressionassessment–TGF-betapathwaygeneswithsignificantlyhigherexpressioninnormoxiagroup Genesymbol Genename Foldchange(Normoxia/Hyperoxia) Comparisonofexpressionbetweengroups; t-test(pvalue) Crebbp CREBBindingProtein 1.95 0.013 Smad1 SmadFamilyMember1 1.43 0.04 Smad2 SmadFamilyMember2 1.31 0.045 Smad3 SmadFamilyMember3 1.2 0.038 Acvr1c ActivinAReceptorType1C 1.51 0.041 M.Bik-Multanowskietal. Free Radical Biology and Medicine 121 (2018) 51–56 54 M. Bik-Multanowski et al. Fig. 4. TGF-β pathway genes exhibiting different methylation patterns. regulating these cellular activities [23,24]. Type I and type II receptors for TGF-β are serine/threonine kinases that signal through the SMAD family of transcriptional regulators. Temporal fluctuations in expres sion of both receptors were demonstrated in rats undergoing prolonged exposure to 100% oxygen [24]. Crebbp and Creb1 genes play critical roles in embryonic lung de velopment [25], growth control, and homeostasis by coupling chro matin remodeling to transcription factor recognition. Crebbp mediates cAMP-gene regulation by binding specifically to phosphorylated CREB1 protein, which is an important regulator of apoptosis in oxidant-medi ated responses of lung epithelial cells [26]. cAMP response element binding protein 1 (Creb1) activity is crucial for the development and differentiation of the conducting and distal lung epithelium. The Creb1 transcription factor regulates cellular gene expression in response to elevated levels of intracellular cAMP. It is noteworthy that TGF-β re presses the cyclin A gene through a cyclic AMP (cAMP) response ele ment, which complexes with the cAMP response element binding pro tein (Creb1) [27]. Wedetected fluctuations in the methylation of the genes mentioned above in response to hyperoxia. The findings indicate a role for these genes in the pathogenesis of lung injury, such as in BPD. We also de monstrated complementary expression changes of some important genes from the TGF-β pathway and observed altered expression of several genes involved in apoptosis regulation and in the cellular re sponse to oxidative stress. However, DNA methylation is probably counter-balanced by other epigenetic processes, which might behave dynamically and in a cell-specific manner. Therefore, the observed differences in expression of selected genes might only partially reflect the hyperoxia-related methylation pattern. Animportant limitation of the study lies in the fact that the presence or absence of long-term alteration in methylation and protein synthesis may depend on the duration of hyperoxia. We exposed the animals to hyperoxia for a relatively long time. We induced alveolar development arrest and could observe morphological abnormalities in lung tissue, which resemble those seen in BPD. However, we were not able to assess potential short-term fluctuations of the DNA methylation with resulting alterations of genome expression. Detailed investigation of the dy namics of TGF-β pathway expression alterations would probably re quire repeated assessment of expression changes in lung tissue in a larger study. Also, we propose that the different cell types of the lung should be studied separately. Some limitations of the mouse model in studies on oxygen exposure in neonates should also be mentioned. These include the relatively high resistance of rodents to oxygen as compared to preterm infants [28],as well as problems to reflect in the animal model the gestational age differences or the effects of prolonged mechanical ventilation. In summary, our results suggest the involvement of methylation of TGF-β pathway genes in the reaction of lung tissue to hyperoxia. It seems that the excess of oxygen can trigger hypermethylation of the pathway, with subsequent decrease of activity of pivotal genes and stimulation of apoptosis. This, in turn can result in alteration of lung morphogenesis with disturbed branching and alveolarization. However, additional experiments are necessary to elucidate the precise mechan isms of epigenetic influence of hyperoxia on lung tissue and their contribution to development of BPD. Declaration of interest None. Funding The work wasfunded by South and Eastern Norway Regional Health Authority; Source number: 6051, Project no.: 39570 and, partially, by the Polish-Norwegian Research Program, operated by the National Centre for Research and Development under the Norwegian Financial Mechanism 2009–2014 in the frame of Project Contract no. Pol-Nor/ 196065/54/2013. Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.freeradbiomed.2018.04.

References [1] O.D. Saugstad, Oxygen and oxidative stress in bronchopulmonary dysplasia, J. Perinat. Med. 38 (6) (2010) 571–577. [2] P. Kwinta, M. Bik-Multanowski, Z. Mitkowska, T. Tomasik, M. Legutko, J.J. Pietrzyk, Genetic risk factors of bronchopulmonary dysplasia, Pediatr. Res. 64 M. Bik-Multanowski et al. (6) (2008) 682–688. [3] T.M. Asikainen, N.S. Waleh, B.K. Schneider, R.I. Clyman, C.W. White, Enhancement of angiogenic effectors through hypoxia-inducible factor in preterm primate lung in vivo, Am. J. Physiol. Lung Cell. Mol. Physiol. 291 (4) (2006) L588–L595. [4] J. Gauldie, T. Galt, P. Bonniaud, C. Robbins, M. Kelly, D. Warburton, Transfer of the active form of transforming growth factor-beta 1 gene to newborn rat lung induces changes consistent with bronchopulmonary dysplasia, Am. J. Pathol. 163 (6) (2003) 2575–2584. [5] A. Chetty, S. Andersson, P. Lassus, H.C. Nielsen, Insulin-like growth factor-1 (IGF-1) and IGF-1 receptor (IGF-1R) expression in human lung in RDS and BPD, Pediatr. Pulmonol. 37 (2) (2004) 128–136. [6] V. Valinluck, L.C. Sowers, Inflammation-mediated cytosine damage: a mechanistic link between inflammation and the epigenetic alterations in human cancers, Cancer Res. 67 (12) (2007) 5583–5586. [7] V.V. Lao, J.L. Herring, C.H. Kim, A. Darwanto, U. Soto, L.C. Sowers, Incorporation of 5-chlorocytosine into mammalian DNA results in heritable gene silencing and altered cytosine methylation patterns, Carcinogenesis 30 (5) (2009) 886–893. [8] N. Chia, L. Wang, X. Lu, M.C. Senut, C. Brenner, D.M. Ruden, Hypothesis: en vironmental regulation of 5-hydroxymethylcytosine by oxidative stress, Epigenetics 6 (7) (2011) 853–856. [9] V. Valinluck, L.C. Sowers, Endogenous cytosine damage products alter the site se lectivity of human DNA maintenance methyltransferase DNMT1, Cancer Res. 67 (3) (2007) 946–950. [10] B. Reinhart, J.R. Chaillet, Genomic imprinting: cis-acting sequences and regional control, Int. Rev. Cytol. 243 (2005) 173–213. [11] M.A. Alejandre-Alcazar, G. Kwapiszewska, I. Reiss, O.V. Amarie, L.M. Marsh, J. Sevilla-Perez, M. Wygrecka, B. Eul, S. Kobrich, M. Hesse, R.T. Schermuly, W. Seeger, O. Eickelberg, R.E. Morty, Hyperoxia modulates tgf-beta/bmp signaling in a mouse model of bronchopulmonary dysplasia, Am. J. Physiol. Lung Cell. Mol. Physiol. 292 (2007) L537–L549. [12] W. Huang da, B.T. Sherman, R.A. Lempicki, Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources, Nat. Protoc. 4 (1) (2009) 44–57. [13] M. Kanehisa, M. Furumichi, M. Tanabe, Y. Sato, K. Morishima, KEGG: new per spectives on genomes, pathways, diseases and drugs, Nucleic Acids Res. 45 (D1) (2017) D353–D361. [14] E.J. Wollen, Y. Sejersted, M.S. Wright, M. Bik-Multanowski, A. Madetko-Talowska, C.C. Günther, S. Nygård, P. Kwinta, J.J. Pietrzyk, O.D. Saugstad, Transcriptome profiling of the newborn mouse lung after hypoxia and reoxygenation: hyperoxic reoxygenation affects mTOR signaling pathway, DNA repair, and JNK-pathway regulation, Pediatr. Res. 74 (5) (2013) 536–544. [15] G.T. Wagenaar, S.A. ter Horst, M.A. van Gastelen, L.M. Leijser, T. Mauad, P.A. van der Velden, E. de Heer, P.S. Hiemstra, B.J. Poorthuis, F.J. Walther, Gene expression profile and histopathology of experimental bronchopulmonary dysplasia induced by prolonged oxidative stress, Free Radic. Biol. Med. 36 (6) (2004) 782–801. [16] B.H. Munkeby, W.B. Børke, K. Bjørnland, L.I. Sikkeland, G.I. Borge, J. Lømo, S. Rivera, M. Khrestchatisky, B. Halvorsen, O.D. Saugstad, Resuscitation of hypoxic piglets with 100% O2 increases pulmonary metalloproteinases and IL-8, Pediatr. Res. 58 (3) (2005) 542–548. [17] P.J. Lee, A.M. Choi, Pathways of cell signaling in hyperoxia, Free Radic. Biol. Med. 35 (4) (2003) 341–350. [18] J. Viña, M. Vento, F. García-Sala, I.R. Puertes, E. Gascó, J. Sastre, M. Asensi, F.V. Pallardó, L-cysteine and glutathione metabolism are impaired in premature infants due to cystathionase deficiency, Am. J. Clin. Nutr. 61 (5) (1995) 1067–1069 (Erratum in: Am. J. Clin. Nutr. 89(6) (2009) 1951). [19] B.C. Schock, D.G. Sweet, H.L. Halliday, I.S. Young, M. Ennis, Oxidative stress in lavage fluid of preterm infants at risk of chronic lung disease, Am. J. Physiol. Lung Cell. Mol. Physiol. 281 (6) (2001) L1386–L1391. [20] C.M. Chen, Y.C. Liu, Y.J. Chen, H.C. Chou, Genome-wide analysis of DNA methy lation in hyperoxia-exposed newborn rat lung, Lung 195 (5) (2017) 661–669. [21] X. Gu, J. Sun, S. Li, X. Wu, L. Li, Oxidative stress induces DNA demethylation and histone acetylation in SH-SY5Y cells: potential epigenetic mechanisms in gene transcription in Aβ production, Neurobiol. Aging 34 (4) (2013) 1069–1079. [22] B. Reinhart, J.R. Chaillet, Genomic imprinting: cis-acting sequences and regional control, Int. Rev. Cytol. 243 (2005) 173–213. [23] R.K. Assoian, B.E. Fleurdelys, H.C. Stevenson, P.J. Miller, D.K. Madtes, E.W. Raines, R. Ross, M.B. Sporn, Expression and secretion of type beta transforming growth factor by activated human macrophages, Proc. Natl. Acad. Sci. USA 84 (17) (1987) 6020–6024. [24] Y. Zhao, B.J. Gilmore, S.L. Young, Expression of transforming growth factor-beta receptors during hyperoxia-induced lung injury and repair, Am. J. Physiol. 273 (2 Pt 1) (1997) L355–L362. [25] A.D. Bird, S.J. Flecknoe, K.H. Tan, P.F. Olsson, N. Antony, T. Mantamadiotis, R. Mollard, S.B. Hooper, T.J. Cole, cAMP response element binding protein is re quired for differentiation of respiratory epithelium during murine development, PLoS One 6 (3) (2011) e17843. [26] C.A. Barlow, A. Shukla, B.T. Mossman, K.M. Lounsbury, Oxidant-mediated cAMP response element binding protein activation: calcium regulation and role in apop tosis of lung epithelial cells, Am. J. Respir. Cell Mol. Biol. 34 (1) (2006) 7–14. [27] K. Kamiya, K. Sakakibara, E.J. Ryer, R.P. Hom, E.B. Leof, K.C. Kent, B. Liu, Phosphorylation of the cyclic AMP response element binding protein mediates transforming growth factor beta-induced downregulation of cyclin A in vascular smooth muscle cells, Mol. Cell. Biol. 27 (9) (2007) 3489–3498. [28] Y. Chen, P.L. Whitney, L. Frank, Comparative responses of premature versus full term newborn rats to prolonged hyperoxia, Pediatr. Res. 35 (2) (1994) 233–237.