Interleukin-1 receptor antagonist prevents murinebronchopulmonary dysplasia induced by perinatalinflammation and hyperoxiaMarcel F. Nolda, Niamh E. Manganb,1, Ina Rudloffa,1, Steven X. Choa, Nikeh Shariatiana, Thilini D. Samarasinghea,Elizabeth M. Skuzaa, John Pedersenc, Alex Veldmana, Philip J. Bergera, and Claudia A. Nold-Petrya,2

aRitchie Centre and bCentre for Innate Immunity and Infectious Diseases, Monash Institute of Medical Research, Monash University, Clayton, Melbourne,VIC 3168, Australia; and cTissuPath, Mount Waverley, VIC 3149, Australia

Edited* by Charles A. Dinarello, University of Colorado Denver, Aurora, CO, and approved July 22, 2013 (received for review April 27, 2013)

Bronchopulmonary dysplasia (BPD) is a common lung disease ofpremature infants, with devastating short- and long-term con-sequences. The pathogenesis of BPD is multifactorial, but alltriggers cause pulmonary inflammation. No therapy exists; there-fore, we investigated whether the anti-inflammatory interleukin-1receptor antagonist (IL-1Ra) prevents murine BPD. We precipitatedBPD by perinatal inflammation (lipopolysaccharide injection topregnant dams) and rearing pups in hyperoxia (65% or 85% O2).Pups were treated daily with IL-1Ra or vehicle for up to 28 d.Vehicle-injected animals in both levels of hyperoxia developeda severe BPD-like lung disease (alveolar number and gas exchangearea decreased by up to 60%, alveolar size increased up to four-fold). IL-1Ra prevented this structural disintegration at 65%, butnot 85% O2. Hyperoxia depleted pulmonary immune cells by 67%;however, extant macrophages and dendritic cells were hyperacti-vated, with CD11b and GR1 (Ly6G/C) highly expressed. IL-1Ra par-tially rescued the immune cell population in hyperoxia (doublingthe viable cells), reduced the percentage that were activated by upto 63%, and abolished the unexpected persistence of IL-1α and IL-1β on day 28 in hyperoxia/vehicle-treated lungs. On day 3, perina-tal inflammation and hyperoxia each triggered a distinct pulmo-nary immune response, with some proinflammatory mediatorsincreasing up to 20-fold and some amenable to partial or completereversal with IL-1Ra. In summary, our analysis reveals a pivotal rolefor IL-1α/β in murine BPD and an involvement for MIP (macro-phage inflammatory protein)-1α and TREM (triggering receptorexpressed on myeloid cells)-1. Because it effectively shields new-born mice from BPD, IL-1Ra emerges as a promising treatment fora currently irremediable disease that may potentially brighten theprognosis of the tiny preterm patients.

anti-inflammatory therapy | cytokines | receptor blockade |neonatal immunity

Modern perinatal medicine has substantially improved sur-vival of extremely premature infants, but at the price of

a steadily rising incidence of a severe lung disease known asbronchopulmonary dysplasia (BPD). BPD is now the most com-mon chronic lung disease in infants and represents a major bur-den for babies, their families, and health services (1). The adverseimpact of BPD on cardiovascular and respiratory health andneurodevelopmental outcome can persist into adulthood (2, 3).In addition, BPD exacts a considerable late mortality, with infantsdying from common and usually harmless viral infections. Accord-ing to National Institutes of Health estimates, treating infantswith BPD in the United States costs $2.4 billion per year (4),making it the second most expensive childhood disease afterasthma. Thus, BPD represents one of the greatest unmet ther-apeutic challenges in neonatology, yet efforts to find a safe andeffective therapy have failed.In the presurfactant era, BPD was largely characterized by

ventilator-induced lung injury and consequent fibrosis, whereasthe “new BPD” features severely reduced alveolarization and

vascularization of the developing lung (5). The pathogenesis ofBPD is multifactorial, resulting from pre- and postnatal infectionand oxygen toxicity. Trauma caused by mechanical ventilationrepresents a third cause of BPD (Fig. S1), but its impact is on thedecline as a result of modern, gentle ventilation strategies. Theseinsults trigger pulmonary inflammation, now recognized as thecommon final pathway and main culprit in the pathogenesis ofBPD. Inflammation leads to decreased alveolar septation, reducedmaturation of epithelial type 1 and type 2 cells, parenchymalthickening, and reduced capillary density (6). The mediators ofthis process include chemokines, adhesion molecules, and proin-flammatory cytokines such as IL-6, TNF, and particularly IL-1β,which have been associated with BPD and/or an increased risk foran adverse clinical course (6–11).To date, the only anti-inflammatory agents tested in BPD are

glucocorticoids, and as expected they reduce its incidence. How-ever, glucocorticoids are used sparingly because their side effectsare severe and include cerebral palsy and inhibition of alveolargrowth, whereby they impede the very pathway essential for per-manent healing of BPD (12).Taking a fresh approach, we set out to test the hypothesis that

IL-1 receptor antagonist (IL-1Ra) is effective in preventing BPD.Our rationale was that (i) IL-1α and IL-1β are already known tobe important malefactors in BPD (6–9, 11, 13–15); (ii) tellingly,a low-bioactivity polymorphism of il1rn is strongly associated(odds ratio 11.7) with BPD (16); and (iii) IL-1Ra features a fa-vorable safety profile that is well-established in a broad spectrumof diseases (17, 18), including neonatal cases (19, 20). Wetherefore embarked on the exploration of IL-1Ra as a candidatetherapy for BPD using a clinically relevant mouse model.

ResultsPerinatal Inflammation and 85% (vol/vol) O2 Produce a Profound andIrreversible BPD-Like Lung Injury. The combination of perinatal in-flammation and postnatal hyperoxia triggers a severe BPD-like lungdisease in rodents (21, 22). We selected this double-hit model as themultifactorial etiology of BPD is taken into account.Perinatal LPS and a fraction of inspired (Fi) O2 of 0.85 in-

duced a severe lung disease in newborn mice. Histologicalanalysis on day 28 demonstrated arrested alveolar development,with a decreased number of large, simplified alveoli and sparsesecondary septation (Fig. 1A). Quantitative analysis confirmedthe histological appearance, revealing a more than 60% reduction

Author contributions: M.F.N., A.V., P.J.B., and C.A.N.-P. designed research; M.F.N., N.E.M.,I.R., S.X.C., N.S., T.D.S., E.M.S., J.P., P.J.B., and C.A.N.-P. performed research; M.F.N., N.E.M.,A.V., P.J.B., and C.A.N.-P. analyzed data; M.F.N., A.V., P.J.B., and C.A.N.-P. wrotethe paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1N.E.M. and I.R. contributed equally to this work.2To whom correspondence should be addressed. E-mail: claudia.nold@monash.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1306859110/-/DCSupplemental.

14384–14389 | PNAS | August 27, 2013 | vol. 110 | no. 35 www.pnas.org/cgi/doi/10.1073/pnas.1306859110

Dow

nloa

ded

by g

uest

on

Oct

ober

25,

202

0

in alveolar number (Fig. 1B), an almost fourfold increase in al-veolar size (Fig. 1C), and a 56% decrease in the surface area-to-volume ratio (SVR), a surrogate parameter for gas exchange(Fig. 1D). Newborn mice in the hyperoxia groups also exhibitedmarked growth restriction, which may reflect metabolic con-sequences of poor gas exchange. In fact, 85% O2 for 28 d severelyaffected the pups, with only 20% surviving to day 60 comparedwith 100% in the room air group. Of note, we also investigatedthe effects of 85% O2 without antenatal exposure to LPS, butfound lung injury to be inconsistent.Although treatment with IL-1Ra blunted the rise on day 3 in

pulmonary IL-6 and macrophage inflammatory protein (MIP)-2that is triggered by perinatal inflammation alone (Fig. 2, whitebars) and by combined antenatal LPS and postnatal hyperoxia(gray bars), the increases in IL-1α and IL-1β were not affected.Dysalveolarization persisted in all hyperoxia groups, whether ornot they were treated with IL-1Ra (Fig. 1; lower magnificationimages in Fig. S2).Thus, perinatal inflammation and postnatal exposure to 85%

O2 produced an irreparable BPD-like lung disease not amenableto prophylaxis with IL-1Ra, although we observed inhibition ofthe proinflammatory cytokines IL-6 and MIP-2, but not IL-1αand IL-1β, in the lungs on day 3 of life.

Murine BPD Induced by Perinatal Inflammation and Postnatal 65% O2

Is Prevented by IL-1Ra. The irreversible lung damage of 85% O2prompted us to ask whether the hyperoxic insult overwhelmedthe prophylactic or treatment power of IL-1Ra. We thus repeated

the experiments using 65% O2, a level that better reflects themodern clinical setting in which a high FiO2 is avoided wheneverpossible. Quantitative analysis of lung histology showed that thedamage induced by 65% O2 in vehicle-treated pups was milderthan that caused by 85% O2, with the alveolar number reduced by44%, the alveolar size increased 2.3-fold, and the SVR reducedby 35% (Fig. 3 B–D; whole lung sections in Fig. S3).The effects of IL-1Ra were remarkable at an FiO2 of 0.65.

Whereas pups in the hyperoxia-vehicle group displayed a grosslyabnormal, dysalveolarized lung histology (Fig. 3A, Lower Left)on day 28, treatment with IL-1Ra resulted in lungs almost in-distinguishable from those of pups reared in room air (Fig. 3A,Lower Right). IL-1Ra completely prevented the hyperoxia/peri-natal LPS-induced decrease in alveolar number (Fig. 3B) andincrease in alveolar size (Fig. 3C). The SVR, corresponding tothe surface area available for gas exchange, was also improveddramatically by IL-1Ra (Fig. 3D). The growth restriction ob-served with 85% O2 was present at 65%, but to a milder extent,and there was a trend toward improved growth with IL-1Ratreatment (Fig. 3E). When the lungs of pups reared in similarconditions for 14 d were compared structurally, the findings weresimilar to those in pups at 28 d, but the changes were less pro-nounced. Of note, there being no difference between them, datafor the three room air groups were combined for calculatingstructural and molecular parameters; therefore, these groups areidentical in Figs. 1–4. Hence, we demonstrated that IL-1Raprevents murine BPD when the “second hit” of postnatalhyperoxia is milder.

Favorable Modification of the Pulmonary Immune Response Inducedby Antenatal LPS and 65% O2 by IL-1Ra: Molecular Aspects. Theantecedents of BPD, whether oxygen toxicity, fetal and postnatalinflammation, or damaging mechanical ventilation, wreak theirdamage via the common pathway of inflammation (Fig. S1);thus, we next assessed the pulmonary immune responses toperinatal inflammation and hyperoxia in newborn mice on day3 by analysis of mediators of inflammation and on day 28 byimmunohistochemistry (IHC).As demonstrated in Fig. 4 A–E, multiplex protein analysis

revealed that the regulation of cytokines by antenatal LPS andhyperoxia is far from homogenous. Perinatal inflammation alonestrongly increased (up to 17-fold) the abundance of MIP-1β,MIP-2, MCP-5, TNF, IL-1α, IL-6, IL-1Ra, ICAM-1, C5a, trig-gering receptor expressed on myeloid cells-1 (TREM)-1, andM-CSF; conversely, other mediators such as B-lymphocyte che-moattractant (BLC), cytokine-induced neutrophil chemoattractant(KC), IL-10, MIP-1α, and IL-1β exhibited little change. The levels

Fig. 1. Lung histology in 28-d-old pups exposed to antenatal LPS andpostnatal hyperoxia at 85%. Pregnant dams were injected with 150 μg/kgLPS at day 14 of gestation. After delivery, newborn mice were exposed to anFiO2 of 0.21 (room air) or 0.85 and received daily s.c. injections of eithervehicle or IL-1Ra (10 mg/kg); n = 10–27 per group. Another group of damsreceived vehicle instead of LPS antenatally; their pups breathed room air andwere injected with vehicle (no antenatal LPS, air vehicle; n = 6). Lungs wereassessed after 28 d. (A) One representative 500 μm × 500 μm slide per groupis depicted. Scale bars: 100 μm. (B–D) Scans of whole lungs were analyzed forthe number of alveoli per square millimeter (B), the size of the alveoli (C),and the SVR (D). Data are shown as means ± SEM. ♦P < 0.05; ♦♦♦P < 0.001 forno antenatal LPS air vehicle vs. hyperoxia-vehicle; ❖P < 0.05 and ❖❖❖P < 0.001for room air vehicle vs. hyperoxia-vehicle; **P < 0.01, room air vehicle vs.room air IL-1Ra.

Fig. 2. Abundance of pulmonary cytokines on day 3 after perinatal in-flammation and 85% hyperoxia. Following antenatal LPS or vehicle, 3 d of21% (room air) or 85% O2 and daily postnatal s.c. injection with IL-1Ra orvehicle, cytokines were determined in the lungs by ELISA. n = 7–20 per group.Data are means of cytokine abundance normalized to total protein (t.p.) ±SEM. □P < 0.05 for no LPS room air vs. room air vehicle; *P < 0.05 for room airvehicle vs. room air IL-1Ra; #P < 0.05 for hyperoxia vehicle vs. hyperoxia IL-1Ra; ♦P < 0.05 for no antenatal LPS air vehicle vs. hyperoxia vehicle.

Nold et al. PNAS | August 27, 2013 | vol. 110 | no. 35 | 14385

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Oct

ober

25,

202

0

of some of these mediators were further increased by the additionof 65% O2, but others were unchanged or even lower. Comparingair vehicle with hyperoxia vehicle, we observed a marked increasein IL-1β (20-fold), BLC (sevenfold), and MIP-1α (sixfold);a moderate increase in KC (threefold) and TREM-1 (2.5-fold);a decrease in TNF (by 48%) and M-CSF (by 42%); and a minorincrease (<twofold) or no change in the other mediators.With regard to response to treatment with IL-1Ra, we found

a more than 50% reduction in the abundance of IL-1β (94%),IL-6 (91%), IL-1α (90%), TREM-1 (85%), MIP-2 (75%), MIP-1α (72%), TNF (70%), KC (69%), MIP-1β (56%), and BLC(50%) in the hyperoxia-exposed animals. The only cytokines thatwere significantly changed by IL-1Ra in room air were IL-6,MIP-2, and TNF (reduced by 82%, 82%, and 58%, respectively).To assess the abundance of IL-1α and IL-1β in the lungs on

day 28, we performed IHC analysis. As expected, the lungs of theroom air groups were virtually devoid of IL-1α and IL-1β (Fig. 4Fand Fig. S4). Unexpectedly, IL-1β was readily detectable in thelungs of the hyperoxia vehicle animals, but its abundance inanimals treated with IL-1Ra was reduced to levels nearly iden-tical to those in the room air groups (Fig. 4F). The increase inIL-1α induced by hyperoxia was less pronounced in comparisonwith IL-1β, but IL-1α also reverted to background with IL-1Ratreatment (Fig. S4).In summary, perinatal LPS and postnatal hyperoxia produce

distinct molecular signatures regarding the pulmonary immuneresponse on day 3 of life, which can in part be modified by IL-1receptor blockade. On day 28, IL-1Ra abolishes the persistenceof IL-1 in lung tissue exposed to hyperoxia.

Effects of IL-1Ra on Cellular Aspects of the Immune Response toPerinatal LPS and 65% O2. Neutrophils, lymphocytes, monocytes/macrophages, natural killer (NK) cells, and dendritic cells (DC)were enumerated and evaluated for their level of activation onday 28. The overall number of immune cells was approximatelythree times lower in the hyperoxia groups at 65% and 85% thanin the room air groups (e.g., 0.45 vs. 1.4 × 106 cells at 65%; Fig.5A and Fig. S5A). IL-1Ra conferred a partial protection at 65%(hyperoxia vehicle 0.45 vs. hyperoxia IL-1Ra 0.85 × 106 cells, P <0.05). There was no difference in the percentage population ofCD4+ and CD8+ T cells, B cells, conventional and plasmacytoidDC, or F4/80+ macrophages between any of the groups among20,000 acquired CD45+ cells (Fig. 5 B and C); however, hyper-oxia caused an increase in the activation of macrophages and DC(Fig. 5 D–G). This activation was demonstrated by assessing thesurface expression of CD11b and GR1 (Ly6G/C), which wereboth increased in the hyperoxia vehicle animals compared withall pups reared in room air. Treatment with IL-1Ra reduced thepercentage of CD11b+ and GR1+ cells. At 85% O2, the in-flammatory phenotype of the pulmonary monocytes and DC wasmore pronounced than at 65% (Fig. S5 B and C). In addition, weobserved a marked increase in a population of CD11b−GR1hi

granulocytes that was reduced by IL-1Ra (Fig. S5).Similar studies were performed on day 14. Most of the findings

shown in Fig. 5 and Fig. S5 for day 28 were detectable, but thechanges were less pronounced and not statistically significant.Taken together, these results demonstrate that perinatal in-

flammation plus postnatal hyperoxia induce a reduction in totalimmune cells in the lung and a marked activation of theremaining cells, particularly macrophages and DC. These effectscan be ameliorated by IL-1Ra.

DiscussionThe central role of perinatal inflammation and its possible con-sequence, the fetal inflammatory response syndrome, led the BPDGroup of the American Academy of Pediatrics (AAP) to identifythe development of anti-inflammatory agents as a research priority(23). By using a clinically relevant “double-hit” model of BPD thatfeatures the two threats commonly experienced by the neonatallung in the age of gentle ventilation—perinatal inflammation andoxygen toxicity—we demonstrate that prophylactic administrationof IL-1Ra can prevent this devastating lung disease in newbornmice. Importantly, we show the clear concentration-dependentnature of oxygen toxicity because IL-1Ra treatment effectivelyprotected the lungs in 65% but not 85% O2. This finding providesa strong basis for the ongoing efforts to limit the O2 level to whichbabies are exposed (24). Beyond responding to the AAP’s call foraction to reduce the burden of BPD, we also shed light on theneglected field of neonatal immunology.We demonstrate that a moderate dose of maternal LPS sub-

stantially increases the abundance of a group of proinflammatorymediators in the lungs of newborn pups; nevertheless, the lungsdevelop normally without a second insult. In combination withhyperoxia, lung inflammation led to a severe BPD-like lung diseasethat was preventable by attenuating this immune response. Theseresults emphasize that the etiology of BPD is indeed multifactorial;perinatal inflammation (e.g., deriving from chorioamnionitis)contributes to BPD provided other events occur. A recent reviewdiscussed this controversial concept (24), citing the finding ofa decreased risk for BPD in infants born after chorioamnionitisisolated to the placenta and chorion, but an increased risk for BPDonce the fetus itself was affected (25). The newborn mice in ourmodel can be regarded as belonging to the latter category.Neonatal immunology is a poorly researched field, with only

a handful of studies devoted to the pulmonary immune response;thus, our report considerably expands knowledge in this field. Adecreased proliferation of structural lung cells in response tohyperoxia has been described (14); we now report a significantloss of immune cells that was partially rescued by IL-1Ra. An-other enlightening result was that activation of macrophages andDC emerged as a hallmark of murine BPD, a finding that accords

Fig. 3. Lung histology at day 28 of life after perinatal LPS or vehicle and65% O2. Animals were treated identically to those in Fig. 1 except for re-duction of the FiO2 to 0.65. Room air groups from each of the 28-d experi-ments were pooled, so the data are identical to those in Fig. 1. n = 5–27 pergroup. Lungs were collected and analyzed on day 28. (A) One representative500 μm × 500 μm slide for each treatment group is presented. Scale bars:100 μm. (B–D) Computerized quantification of the slides. Alveolar numberper square millimeter (B), alveolar size (C), and SVR (D) are depicted. Dataare means ± SEM. ♦P < 0.05 and ♦♦P < 0.01 for no antenatal LPS air vehicle vs.hyperoxia vehicle; ❖P < 0.05 and ❖❖P < 0.01 for room air vehicle vs. hyperoxiavehicle; **P < 0.01 for room air vehicle vs. room air IL-1Ra; #P < 0.05 and ##P <0.01 for hyperoxia vehicle vs. hyperoxia IL-1Ra. (E) Photograph of one repre-sentative mouse from the hyperoxia vehicle (Left) and the hyperoxiaIL-1Ra groups (Right) on day 28.

14386 | www.pnas.org/cgi/doi/10.1073/pnas.1306859110 Nold et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

25,

202

0

with a 1988 study that reported increased activation of alveolarmacrophages in human infants with BPD compared with thosewithout (26). In our mouse pups, this activation was so pro-nounced that despite the reduction in cell numbers, the abun-dance of most inflammatory mediators was greater in hyperoxiathan in room air. IL-1Ra ameliorated this inflammatory responsein macrophages and DC, as demonstrated by the reduction ofCD11b and GR1. On the molecular level, protein array analysisof the lungs revealed further interesting details about the neo-natal immune response to perinatal inflammation and hyperoxia,each of which triggered a distinct cytokine response.Because perinatal inflammation alone did not lead to mu-

rine BPD, we infer that the mediators most relevant to itsdevelopment are those that exhibit the strongest induction byhyperoxia, namely IL-1β, TREM-1, and the chemokines MIP-1α,BLC, and KC. Because IL-1Ra prevented BPD and markedlyreduced the abundance of IL-1β, TREM-1, and MIP-1α, thesemediators emerge as major culprits of disease initiation and ag-gravation in murine BPD. IL-1α, IL-6, TNF, and MIP-2 were notincreased by hyperoxia on day 3, but still were reduced by anti-inflammatory treatment and thus are likely also involved. An-other piece of evidence for a pivotal role of IL-1 in our model isthat IL-1Ra failed to ameliorate the increases in IL-1α and IL-1β as well as structural lung injury induced by 85% O2, whereas

at 65% O2, the almost complete protection of lung structure wasaccompanied by a reduction of IL-1α and IL-1β to steady-statelevels on days 3 and 28.Although the involvement of IL-1β (6–9, 11, 16) and IL-6 (6, 7,

10) in human BPD is known, literature is sparse on BPD andMIP-1α, MIP-2, and TREM-1. Besides other mononuclear cells,MIP-1α acts on monocytes and DC (27) and may therefore beinvolved in recruiting these cells into the lungs of the newbornmice. Consistently, antibodies to MIP-1α inhibited the harmfulangiogenesis induced by intraamniotic injection of LPS in mice(28). Blockade of MIP-2, a neutrophil chemoattractant, reduceshyperoxia-induced alveolar septal thickening in a rat model ofBPD (29). Increased serum levels of TREM-1, whose mainfunction is to amplify inflammation, are present in pretermbabies, although the association with BPD or death failed toreach significance (P = 0.15) (7). In view of evidence thatTREM-1 facilitates neutrophil migration across airway epithelialcells in a model of pneumonia (30), our report on pulmonaryabundance of TREM-1 may provide a more direct view of itsinvolvement in BPD, albeit in mice.The call by the AAP for research into anti-inflammatory

strategies against BPD led to studies showing that sildenafilprovides benefit, but less than we observed for IL-1Ra (31), andrevealed some promise for bone marrow–derived mesenchymal

Fig. 4. Exploration of mediators of inflammation on days 3 and 28 of the murine BPD model. (A–E) Following the same experimental protocol as animals inFig. 3, cytokine abundance was determined in the lungs obtained on day 3. Room air groups from each of the 3-d experiments were pooled and are thereforeidentical to those in Fig. 2. (C and E) IL-6 and MIP-2 were measured by ELISA. n = 8–20 per group. Graphs show means of cytokine abundance normalized tot.p. ± SEM. (A, B, D) Semiquantitative protein analysis of cytokines (A), other mediators (B), and chemokines (D) was performed by multiplex immunoblottingon the lungs of three animals from each group. Data are plotted as OD normalized to the positive control spots on each membrane in arbitrary units ± SEM.(A–E) □P < 0.05 and □□P < 0.01 for no antenatal LPS air vehicle vs. room air vehicle; *P < 0.05 for room air vehicle vs. room air IL-1Ra; ♦P < 0.05, ♦♦P < 0.01, and♦♦♦P < 0.001 for no antenatal LPS air vehicle vs. hyperoxia vehicle; ❖P < 0.05 and ❖❖P < 0.01 for room air vehicle vs. hyperoxia vehicle; #P < 0.05 for hyperoxiavehicle vs. hyperoxia IL-1Ra. (F) Sections of the same day 28 lungs used to generate Fig. 3 were subjected to IHC assessment of the abundance of pulmonary IL-1β. One representative image for each group is shown; n = 4–7 per group. Scale bars: 100 μm. IL-1α was also determined and is shown in Fig. S4.

Nold et al. PNAS | August 27, 2013 | vol. 110 | no. 35 | 14387

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Oct

ober

25,

202

0

stem cells (32) and blockade of connective tissue growth factor(33) or of TGF-β (34). However, clinical utility for the inter-ventions against the two cytokines is questionable because theyregulate a number of important pathways and interference willentail a plethora of undesired effects. For stem cells, criticalquestions regarding the mechanism of action and safety remainto be addressed.We believe that the data presented here provide a sound

justification for clinical trials that aim to establish IL-1Ra asa safe and effective therapy for BPD. With a view to such clinicaltrials, it is reassuring that IL-1Ra did not adversely affect lung

maturation; in fact, we observed no adverse event whatsoeverwith IL-1Ra in the pups, consistent with the experience ofclinicians that the safety profile of IL-1Ra is superior to that ofother immunomodulating agents such as anti-TNF therapies inadult and pediatric diseases (17, 18). It is relevant to note thatclinical experience with IL-1Ra in pediatric disease abounds,mainly because auto-inflammatory diseases such as familialMediterranean fever, hyper-IgD syndrome, cryopyrin-associatedperiodic syndrome, and systemic juvenile idiopathic arthritis usuallypresent in childhood. Neonatal-onset multisystem inflammatorydisease (NOMID) and deficiency of IL-1Ra (a loss-of-functionmutation in il1rn) both present in early infancy and are suc-cessfully and safely treated with IL-1Ra (17–20). Importantly,a study in 71 preterm infants identified a low-expression il1rngenotype to be strongly associated with an increased suscepti-bility to BPD (odds ratio 11.7 (16)).In addition to these promising prerequisites, IL-1α and IL-1β,

both targets of IL-1Ra, have been shown to play an importantand deleterious role in BPD pathogenesis. Several studiesidentified an association between elevated levels of IL-1β anddevelopment of BPD or an adverse clinical outcome (6, 9). Animbalance between IL-1β and IL-1Ra contributes to prolongedinflammation in BPD (11), and treatment with IL-1Ra protectsfrom loss of function and structural damage in an ozone modelof lung disease (35). Neonatal mice in which IL-1β was condi-tionally overexpressed in airway epithelial cells suffered fromBPD-like disrupted alveolar septation and capillary growth andfeatured elevated levels of MIP-2 (8)–findings consistent withour data. This same study also reported that IL-1β reduced theproduction of VEGF, a mediator crucial for normal neonatalalveolarization (36). Hyperoxia-induced BPD has been associ-ated with an increase in IL-1α and MIP-1α in mice (14), whereasin preterm lambs intratracheal IL-1α triggered severe lung in-flammation (13). In another ovine model, IL-1Ra reduced thefetal pulmonary and systemic inflammatory response to intra-amniotic LPS (15). On a contradictory note, IL-1Ra has beentested unsuccessfully in a model of lung injury induced by hightidal volume ventilation in preterm sheep (37). However, thetrigger for lung damage and the time points investigated weredifferent from our study, which likely explains the absence ofefficacy of IL-1Ra.In summary, we demonstrate that prophylactic administration

of IL-1Ra affords nearly complete protection against BPD andallows almost normal lung development in a clinically relevantmouse model. We also provide intriguing insights into neonatalpulmonary immunology, including the identification of macro-phages and DC as well as IL-1α, IL-1β, MIP-1α, and TREM-1 asthe major cellular and molecular perpetrators of pulmonaryinflammation. Earlier work has shown that IL-1Ra is safe andefficacious in the treatment of human diseases, including inthe pediatric setting. We therefore propose that clinicians andresearchers would be justified in proceeding with a clinical trial ofa nonsteroidal anti-inflammatory drug to fundamentally changethe prognosis of premature infants by preventing the currentlyirremediable BPD, a disease with devastating short- and long-term consequences, including significant mortality.

Materials and MethodsMurine Model of BPD. Based on previous publications (21, 22, 38), pregnantC57BL/6J dams received an i.p. injection of 150 μg/kg of LPS or volume-matched vehicle (normal saline) at 14 d gestation. Within 24 h after birth,pups and dams were randomized to a chamber through which we passedgas at 10 L/min with an FiO2 of 0.21 (room air), 0.65, or 0.85 for a total of 3,14, or 28 d. Pups were also randomly allocated to receive daily s.c. injectionsof IL-1Ra (10 mg/kg) [a moderate dose considering (i) the Food and DrugAdministration recommendation to use 12-fold higher drug doses in murinetrials to compensate for the species-specific differences in activity and (ii) thedose of 1–5 mg/kg in human infants with NOMID (19)] or a similar volume ofvehicle. Temperature (22 °C) and humidity (50–60%) were kept constant.Light was cycled in a 12-h day/night rhythm. Dams were rotated betweenthe groups in a 3-d cycle to limit dam-effects on study outcome and toprotect dams from prolonged hyperoxia. At 3, 14, or 28 d, mice were

Fig. 5. Flow cytometric analysis of the lungs on day 28. The left lungs of theanimals shown in Fig. 3 (perinatal LPS, 65% hyperoxia) were analyzed by flowcytometry. Among viable cells, all immune cells (A) or 20,000 CD45+ cells (B–G)were gated. The graphs show data from one experiment (n = 3–4 per group);a second experiment produced similar results (n = 3–4 per group). (A) Thetotal number of cells isolated from the lungs is depicted. (B and C) Enumer-ation of CD4+ and CD8+ T cells, B220+CD11c− B cells, B220+CD11c+ conven-tional DC, and B220−CD11c+ plasmacytoid DC is plotted as percent of CD45+

cells. (D and E) CD11c+CD11b+-activated monocytes are shown as percent ofCD45+ cells. (E) One exemplary histogram per group of the CD11b stain isdepicted; red and blue line, room air groups; orange line, hyperoxia vehicle;green line, hyperoxia IL-1Ra. (F andG) Assessment of CD11b and GR1, markersof macrophage and granulocyte activation, on CD45+ cells. (F) Analysis of oneexperiment (n = 4 per group); (G) exemplary dot plots of one lung from eachgroup. *P < 0.05, **P < 0.01, and ***P < 0.001 for all comparisons.

14388 | www.pnas.org/cgi/doi/10.1073/pnas.1306859110 Nold et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

25,

202

0

anesthetized with isoflurane, then humanely killed by cervical dislocation.Lungs were processed for histology, flow cytometry, or cytokine assay.

Lung Preparation, Histology, and IHC. After cervical dislocation, 14- and 28-d oldmice were intubated via the trachea and the left lung lobe was tied off at themain bronchus and removed for flow cytometry or cytokine analysis. The rightlung was fixed with 4% paraformaldehyde (PFA) (pH 7.4, instilled at a pressureof 20 cmH2O), removed from the thorax, kept in 4% PFA for a minimum of 2 h,then immediately processed for paraffin embedding and sectioning. Lungtissue was cut into 3-μm sections at three levels from the apex to base of thelung and H&E-stained for histology. For IHC, 4-μm sections were cut anddeparaffinized. Antigen retrieval was performed in 10 mM citrate buffer, pH6.0, in a pressure cooker for 10 min. Endogenous peroxidase activity wasinhibited using a 0.5% H2O2/methanol solution applied to the slides for 15 minfollowed by a 30-min blocking step using 1.5% rabbit serum in PBS (part of theVectastain Elite ABC Kit, Vector Laboratories). Slides were then incubated withgoat anti-mouse IL-1α or IL-1β antibodies (both 1:100, R&D Systems) for 1 h atroom temperature. Secondary antibody incubation and diaminobenzidine(Invitrogen) staining were performed according to the instructions of theVectastain kit. Sections were counterstained with Harris hematoxylin (AmberScientific) and scanned on an Aperio Scanscope (ePathology Solutions).

Cytokine Analysis. At day 3, the entire lung was removed, washed in ice-coldsterile saline, snap frozen in liquid nitrogen, and stored at −80 °C. Foranalysis, the lungs were homogenized in lysis buffer (39) using an UltraTurrax. The homogenate was centrifuged for 10 min at 14,000 × g and thesupernatant was assayed for protein and cytokines. Changes in cytokineabundance resulting from tissue rarefaction, reduced growth, or otherconfounding factors were corrected by normalization to total protein con-centration, i.e., by dividing cytokine content by the total protein content.ELISAs (R&D Systems, MIP-2, IL-1β; BD, IL-6; eBioscience, IL-1α) were per-

formed according to the manufacturers’ instructions. Proteome ProfilerArrays (R&D Systems) were used as described previously (40).

Flow Cytometry. For lung cell isolation on days 14 and 28, the left lung wasminced and digested with collagenase D (1 mg/mL) at 37 °C for 30 min. Afterdigestion, lung homogenates were macerated through a 70-μM cell strainer,red blood cells were lysed (BD Pharm Lyse) for 2× 10 min, then centrifuged at400 × g for 10 min at 18 °C. Pellets were washed twice with PBS and cellswere stained and acquired using a FACS Canto II flow cytometer (BD). Fcreceptors were blocked using anti-CD16/CD32 (eBioscience). Cells were de-termined by surface expression of CD11c, CD11b, and GR1 for monocytesand macrophages; B220+CD11c+ for plasmacytoid DC; B220−CD11c+ forconventional DC; F4/80+CD11b+ for macrophages; B220+CD11c− for B cells;DX5+CD3− for NK cells; and CD4 or CD8 for T cells. A total of 20,000 CD45+

cells were analyzed per sample. A forward scatter gate was used to elimi-nate cell debris and gates were set using fluorophore minus one staininggates. Data were analyzed using Flow Jo software. All antibodies wereobtained from BD Pharmingen, except F4/80 (Serotec).

Statistical Analysis, Ethics, and Analysis of Lung Structure. Please see Sup-porting Information.

ACKNOWLEDGMENTS. This study was supported by research grants from theInterleukin Foundation (to M.F.N. and C.A.N.-P.), the National Health andMedical Research Council and the Australian Research Council (to N.E.M.), theWindermere and Marian E. Flack Foundations (to A.V.), Baxter Bioscience (toA.V. and P.J.B.), the Jack Brockhoff Foundation (to M.F.N. and C.A.N.-P.), R&DSystems, and the Victorian Government’s Operational Infrastructure SupportProgram; a Fellowship by the Deutsche Forschungsgemeinschaft (to M.F.N.);and the Larkins Fellowship from Monash University (to M.F.N.).

1. Baraldi E, Carraro S, Filippone M (2009) Bronchopulmonary dysplasia: Definitions andlong-term respiratory outcome. Early Hum Dev 85(10, Suppl):S1–S3.

2. Short EJ, et al. (2007) Developmental sequelae in preterm infants having a diagnosisof bronchopulmonary dysplasia: Analysis using a severity-based classification system.Arch Pediatr Adolesc Med 161(11):1082–1087.

3. Short EJ, et al. (2003) Cognitive and academic consequences of bronchopulmonarydysplasia and very low birth weight: 8-year-old outcomes. Pediatrics 112(5):e359.

4. Walter EC, Ehlenbach WJ, Hotchkin DL, Chien JW, Koepsell TD (2009) Low birthweight and respiratory disease in adulthood: A population-based case-control study.Am J Respir Crit Care Med 180(2):176–180.

5. Jobe AJ (1999) The new BPD: An arrest of lung development. Pediatr Res 46(6):641–643.6. Bose CL, Dammann CE, Laughon MM (2008) BPD and inflammatory biomarkers in the

premature neonate. Arch Dis Child Fetal Neonatal Ed 93(6):F455–F461.7. Ambalavanan N, et al.; Eunice Kennedy Shriver National Institute of Child Health and

Human Development Neonatal Research Network (2009) Cytokines associated withbronchopulmonary dysplasia or death in extremely low birth weight infants. Pedi-atrics 123(4):1132–1141.

8. Bry K, Whitsett JA, Lappalainen U (2007) IL-1beta disrupts postnatal lung morpho-genesis in the mouse. Am J Respir Cell Mol Biol 36(1):32–42.

9. Cayabyab RG, et al. (2003) Interleukin-1beta in the bronchoalveolar lavage fluid ofpremature neonates: A marker for maternal chorioamnionitis and predictor of ad-verse neonatal outcome. J Matern Fetal Neonatal Med 14(3):205–211.

10. Jónsson B, Tullus K, Brauner A, Lu Y, Noack G (1997) Early increase of TNF alpha andIL-6 in tracheobronchial aspirate fluid indicator of subsequent chronic lung disease inpreterm infants. Arch Dis Child Fetal Neonatal Ed 77(3):F198–F201.

11. Rindfleisch MS, Hasday JD, Taciak V, Broderick K, Viscardi RM (1996) Potential role ofIL-1 in the development of BPD. J Interferon Cytokine Res 16(5):365–373.

12. Halliday H, Ehrenkranz R, Doyle L (2009) Early (< 8d) postnatal corticosteroids forpreventing BPD in preterm infants. Cochrane Database Syst Rev (1):CD001146.

13. Mulrooney N, Jobe AH, Ikegami M (2004) Lung inflammatory responses to intra-tracheal interleukin-1alpha in ventilated preterm lambs. Pediatr Res 55(4):682–687.

14. Warner BB, Stuart LA, Papes RA, Wispé JR (1998) Functional and pathological effectsof prolonged hyperoxia in neonatal mice. Am J Physiol 275(1 Pt 1):L110–L117.

15. Kallapur SG, et al. (2009) IL-1 mediates pulmonary and systemic inflammatory re-sponses to chorioamnionitis induced by lipopolysaccharide. Am J Respir Crit Care Med179(10):955–961.

16. Cakmak BC, et al. (2012) Association between bronchopulmonary dysplasia and MBL2and IL1-RN polymorphisms. Pediatr Int 54(6):863–868.

17. Dinarello CA, Simon A, van der Meer JW (2012) Treating inflammation by blockinginterleukin-1 in a broad spectrum of diseases. Nat Rev Drug Discov 11(8):633–652.

18. Dinarello CA (2011) Interleukin-1 in the pathogenesis and treatment of inflammatorydiseases. Blood 117(14):3720–3732.

19. Rigante D, Leone A, Marrocco R, Laino ME, Stabile A (2011) Long-term response after6-year treatment with anakinra and onset of focal bone erosion in neonatal-onsetmultisystem inflammatory disease (NOMID/CINCA). Rheumatol Int 31(12):1661–1664.

20. Sibley CH, et al. (2012) Sustained response and prevention of damage progressionin patients with neonatal-onset multisystem inflammatory disease treated with

anakinra: A cohort study to determine three- and five-year outcomes. ArthritisRheum 64(7):2375–2386.

21. Choi CW, et al. (2009) BPD in a rat model induced by intra-amniotic inflammation andpostnatal hyperoxia: Morphometric aspects. Pediatr Res 65(3):323–327.

22. Velten M, Heyob KM, Rogers LK, Welty SE (2010) Deficits in lung alveolarization andfunction after systemic maternal inflammation and neonatal hyperoxia exposure.J Appl Physiol 108(5):1347–1356.

23. Walsh MC, et al. (2006) Summary proceedings from the bronchopulmonary dysplasiagroup. Pediatrics 117(3 Pt 2):S52–S56.

24. Jobe AH (2011) The new bronchopulmonary dysplasia. Curr Opin Pediatr 23(2):167–172.25. Been JV, et al. (2009) Histologic chorioamnionitis, fetal involvement, and antenatal ste-

roids: Effects on neonatal outcome in preterm infants.Am J Obstet Gynecol 201(6):e1–e8.26. Clement A, Chadelat K, Sardet A, Grimfeld A, Tournier G (1988) Alveolar macrophage

status in BPD. Pediatr Res 23(5):470–473.27. He S, et al. (2008) MIP-3alpha and MIP-1alpha rapidly mobilize dendritic cell pre-

cursors into the peripheral blood. J Leukoc Biol 84(6):1549–1556.28. Miller JD, Benjamin JT, Kelly DR, Frank DB, Prince LS (2010) Chorioamnionitis stim-

ulates angiogenesis in saccular stage fetal lungs via CC chemokines. Am J Physiol LungCell Mol Physiol 298(5):L637–L645.

29. Deng H, Mason SN, Auten RL, Jr. (2000) Lung inflammation in hyperoxia can beprevented by antichemokine treatment in newborn rats. Am J Respir Crit Care Med162(6):2316–2323.

30. Klesney-Tait J, et al. (2013) Transepithelial migration of neutrophils into the lungrequires TREM-1. J Clin Invest 123(1):138–149.

31. de Visser YP, et al. (2009) Sildenafil attenuates pulmonary inflammation, mortalityand right ventricular hypertrophy in neonatal hyperoxic lung injury. Respir Res 10:30.

32. Zhang X, et al. (2012) Role of bone marrow-derived mesenchymal stem cells in theprevention of hyperoxia-induced lung injury in newborn mice. Cell Biol Int 36(6):589–594.

33. Alapati D, et al. (2011) Connective tissue growth factor antibody therapy attenuateshyperoxia-induced lung injury in neonatal rats.Am J Respir Cell Mol Biol 45(6):1169–1177.

34. Nakanishi H, Sugiura T, Streisand JB, Lonning SM, Roberts JD, Jr. (2007) TGF-beta-neutralizing antibodies improve pulmonary alveologenesis and vasculogenesis in theinjured newborn lung. Am J Physiol Lung Cell Mol Physiol 293(1):L151–L161.

35. Park JW, et al. (2004) Interleukin-1 receptor antagonist attenuates airway hyper-responsiveness following exposure to ozone. Am J Respir Cell Mol Biol 30(6):830–836.

36. Thebaud B, et al. (2005) VEGF gene therapy increases survival, promotes lung an-giogenesis, and prevents alveolar damage in hyperoxia-induced lung injury: Evidencethat angiogenesis participates in alveolarization. Circulation 112(16):2477–2486.

37. Hillman NH, et al. (2010) Inhibitors of inflammation and endogenous surfactant poolsize as modulators of lung injury with initiation of ventilation in preterm sheep.Respir Res 11:151.

38. Alejandre-Alcazar MA, et al. (2007) Hyperoxia modulates TGF-beta/BMP signaling ina mouse model of BPD. Am J Physiol Lung Cell Mol Physiol 292(2):L537–L549.

39. Petry C, Fritz G, Pfeilschifter J, Huwiler A (2004) Inhibition of Rho modulates cytokine-induced PGE2 formation in mesangial cells. Biochim Biophys Acta 1636(2-3):108–118.

40. Nold MF, et al. (2010) IL-37 is a fundamental inhibitor of innate immunity. Nat Immunol11(11):1014–1022.

Nold et al. PNAS | August 27, 2013 | vol. 110 | no. 35 | 14389

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Oct

ober

25,

202

0