Hyperoxic exposure of immature mice increases the inflammatory response to subsequent rhinovirus infection: Association with danger signals

Abstract

Infants with a history of prematurity and bronchopulmonary dysplasia (BPD) have a high risk of asthma and viral-induced exacerbations later in life.

We hypothesized that hyperoxic exposure, a predisposing factor to BPD, modulates the innate immune response, producing an exaggerated pro-inflammatory reaction to viral infection.

Two-to-3 day-old C57BL/6J mice were exposed to air or 75% oxygen for 14 days. Mice were infected intranasally with rhinovirus (RV) immediately after O₂ exposure. Lung mRNA and protein expression, histology, dendritic cells (DCs) and airways responsiveness were assessed 1-12 days after infection. Tracheal aspirates from premature human infants were collected for mRNA detection.

Hyperoxia increased lung IL-12 expression which persisted up to 12 days post-exposure. Hyperoxia-exposed RV-infected mice showed further increases in IL-12 and increased expression of IFN-γ, TNF-α, CCL2, CCL3 and CCL4, as well as increased airway inflammation and responsiveness. In RV-infected, air-exposed mice the response was not significant. Induced IL-12 expression in hyperoxia-exposed, RV-infected mice was associated with increased IL-12-producing CD103+ lung DCs. Hyperoxia also increased expression of Clec9a, a CD103+ DC-specific damaged cell-recognition molecule. Hyperoxia increased levels of ATP metabolites and expression of adenosine receptor A1, further evidence of cell damage and related signaling. In human preterm infants, tracheal aspirate Clec9a expression positively correlated with the level of prematurity.

Hyperoxic exposure increases the activation of CD103+, Clec9a+ DCs, leading to increased inflammation and airway hyperresponsiveness upon RV infection. In premature infants, danger signal-induced DC activation may promote pro-inflammatory airway responses, thereby increasing respiratory morbidity.

INTRODUCTION

Premature birth is associated with the development of recurrent wheezing and asthma (1, 2). Infants with bronchopulmonary dysplasia (BPD), the most common chronic pulmonary disease following premature birth (3), have increased respiratory morbidity and healthcare utilization, including increased medication use, wheezing and hospital readmissions (4–6). Infants with BPD are at particular risk for severe disease and hospitalization following respiratory viral infections including respiratory syncytial virus (RSV) (7) and rhinovirus (RV), the most common childhood viral pathogen (8–11).

The precise mechanisms leading to higher respiratory morbidity and abnormal airway function in prematurely-born children are not well understood. While early reports of lung structure in BPD demonstrated bronchial submucosal fibrosis and increased smooth muscle mass as well as hypoalveolarization and alveolar septal fibrosis (12, 13), lung structural abnormalities in the “new,” surfactant-treated BPD are relatively mild, with hypoalveolarization, little or no alveolar septal fibrosis and an absence of airway structural changes (14–16). At first blush, these structural abnormalities do not seem to completely explain the increased wheezing, airways obstruction, respiratory morbidity or susceptibility to viral infections observed in premature infants.

Instead, it is conceivable that early-life exposures may modulate the immune response, leading to asthma development. For example, neonatal hyperoxia exposure, a model of BPD, enhanced the subsequent inflammatory response to influenza A virus infection in adult mice (17). We hypothesize that neonatal hyperoxic exposure modulates the neonatal innate immune system, producing an exaggerated pro-inflammatory response to viral infection, leading to increased respiratory morbidity and airways hyperresponsiveness. To test this, we examined the effects of hyperoxic exposure on the inflammatory response to infection with RV, the most common respiratory tract infection. We found that hyperoxia increases IL-12 production from lung dendritic cells (DCs), leading to exaggerated pro-inflammatory responses to subsequent RV infection. Hyperoxic exposure also increased expression of the DC-specific damaged cell-recognition molecule Clec9a and the adenosine receptor Adora1, evidence of cell damage-related signaling.

METHODS

Ethics statement

The animal experiments were performed in strict accordance with the NIH Guide for the Care and Use of Laboratory Animals recommendations. The protocol was approved by the University of Michigan Committee on Use and Care of Animals.

Generation of RV

RV1B or RV2 (both from American Type Culture Collection, Manassas, VA) were grown in HeLa cells, concentrated, partially purified, and titered as described (18). Viral titers were measured by plaque assay (19).

Animal model

Two-to-three day-old C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME) were exposed to air or 75% oxygen for up to 14 days using a polypropylene chamber coupled to an oxygen controller and sensor (BioSpherix, Lacona, NY) (20). Dams were exchanged between air and hyperoxia daily. On day 14, hyperoxic exposure was discontinued and mice were treated with either 30 μl of intranasal poly(I:C) (1 μg/μl), RV1B (3 × 10⁸ PFU/ml), or respective control PBS or sham infection (HeLa cell lysate). Mice were sacrificed for analysis 1, 3, 5 or 12 days after poly(I:C) treatment or RV infection. Finally, selected mice were infected with RV 7 days after hyperoxic exposure (day 21 of the protocol), and lungs harvested 2 days after infection.

Quantitative real-time PCR

Mouse whole-lung RNA was prepared using TRIzol (Invitrogen, Carlsbad, CA). Viral RNA and host mRNAs (IL-12p40, IFN-γ, TNF-α, CCL2, CCL3, CCL4, Clec9a, Adora1, Adora2a, Adora2b, Adora3, CD73, CD39, IFN-α, IFN-β, IFN-λ2, IL-4 and IL-13) were quantified using SYBR green technology. The primers used are listed in Table I. The level of gene expression was normalized to mRNA of β-actin using the 2^−ΔCT algorithm. For graphic representation the mRNAs for the target genes relative to β-actin were plotted on the y-axis multiplied by 10⁻³.

Table I.

Primer sequences used for qPCR

Gene	Forward primer (5'–>3')	Reverse primer (5'–>3')
Actb	TGT CGA GTC GCG TCC ACC	TCG TCA TCC ATG GCG AAC TGG
Il12p40	CTC CTG GTT TGC CAT GGT TT	GGG AGT CCA GTC CAC CTC TA
Ifng	TGG CTG TTT CTG GCT GTT AC	TCC ACA TCT ATG CCA CTT GAG TT
Tnfa	ATG CAC CAC CAT CAA GGA CTC AA	ACC ACT CTC CCT TTG CAG AAC TC
Ccl2	GCT CTC TCT TCC TCC ACC AC	GCG TTA ACT GCA TCT GGC T
Ccl3	TGA AAC CAG CAG CCT TTG	TTG GAC CCA GGT CTC TTT G
Ccl4	AGC AAC ACC ATG AAG CTC TG	CCG GGA GGT GTA AGA GAA AC
Clec9a	GGC CTC TCA GAA GTG CCA AT	CCT GGA AGA ACT TGA TGC CCA
Adora1	GTC AAG ATC CCT CTC CGG TAC A	GGC TAT CCA GGC TTG TTC CA
Adora2a	GCC AGA GCA AGA GGC AGG TAT	AGC CCT TTC CTC ACA AGA GC
Adora2b	GAC TCT TCG CCA TCC CCT TT	CTT TAT ACC TGA GCG GGA CGC
Adora3	ACA GTC AGA TAT AGA ACG GTT ACC A	TGA GAG CTC GCT AAG GTT GC
Cd73	CAG CAT TCC TGA AGA TGC GAC C	TCT GGG TGT CTG AGG TTG TTG
Cd39	AGG AAA CAA AAA GCT GCC CC	CGC ATC CAA CAC AAT CCC ATA CTT A
Ifna	CCA TCC CTG TCC TGA GTG	CCA TGC AGC AGA TGA GTC CTT
Ifnb	GAC GGA GAA GAT GCA GAA GAG TTA C	CCA CCC AGT GCT GGA GAA
Ifnl2	GTG GCC CTG ACC CTG AAG	GTG AAT GTG GCT CAG TGT ATG AAG
Il4	GGT CTC AAC CCC CAG CTA GT	GCC GAT GAT CTC TCT CAA GTG AT
Il13	CCT GGC TCT TGC TTG CCT T	GGT CTT GTG TGA TGT TGC TCA
CLEC9A	ATA GCC CAG CAC CAG ACA CT	TGC AAC AAC TTG ACG CCC AA
GAPDH	CGA CCA CTT TGT CAA GCT CA	AGG GGT CTA CAT GGC AAC TG

Measurement of lung cytokines

Whole lung homogenates in PBS were centrifuged and supernatants analyzed for pro-inflammatory cytokines. IL-12p40, IL-12p70 and TNF-α were measured by ELISA (eBioscience, San Diego, CA and R&D Systems, Minneapolis, MN) and CCL2, CCL3, CCL4, CXCL1 were detected by multiplex assay (Bio-Rad, Hercules, CA). Bronchoalveolar lavage (BAL) was performed using 0.5 ml PBS aliquots. BAL fluid was spun for 15 minutes at 1500 g, and the supernatants were analyzed.

Measurement of AMP, ADP, ATP and adenosine

Plasma and BAL fluid were obtained following 14 days of hyperoxic exposure and assayed for AMP, ADP, ATP and adenosine using mass spectroscopy. To stabilize ATP metabolite concentrations a modified stop solution with multiple enzyme inhibitors including 4.15mM Ethylenediaminetetraacetic acid, 5nM nitrobenzylthioinosine, 10μM forskolin, 100μM 3-isobutyl-1-methylxanthine, 40μM dipyridamole, 10μM erythro-9-(2-hydroxy-3-nonyl)adenine and 10μM 5-iodotubericidin in 40mM tricine buffer with 118mM NaCl and 5mM KCl was used (21). BAL was performed using 0.3 ml stop solution aliquots. BALF was centrifuged to remove cells and the supernatant was stored at −80°C. Plasma was obtained by collecting blood by intracardiac puncture and immediately mixing with equal volume stop solution. The samples were centrifuged to separate plasma and stored at −80°C.

Lung histology

Lungs were collected, fixed with 10% formaldehyde (Sigma-Aldrich, St. Louis, MO) and paraffin embedded. Five μm-thick paraffin sections were stained with Hematoxylin and eosin (H&E) (22).

Flow cytometry

Lungs were perfused with PBS containing EDTA (0.5 mM), minced, and digested in collagenase IV (5 mg/ml) and DNase I (23). Cells were filtered and washed with RBC lysis buffer (BD Biosciences, Franklin Lakes NJ) and kept on ice in media containing 10% serum. Dead cells were stained with Pac-Orange Live/Dead fixable dead staining dye (Invitrogen). Lung cells were then stained with fluorescent-labeled antibodies against various leukocyte surface markers (CD45, CD11b, CD11c, F4/80, CD103, MHCII, CD86, Clec9a, CD3ε, TCRβ, CD4, CD69, IFN-γ). Appropriate isotype-matched controls were used in all experiments. Antibodies were purchased from EBiosciences (San Diego, CA) or Biolegend (San Diego, CA). Cells were fixed and analyzed on a Canto2 (Becton-Dickinson, San Jose, CA) or FACSAria II (BD Biosciences) flow cytometer. Results were analyzed using FlowJo software (Tree Star, Ashland, OR). For analysis of intracellular IL-12 or IFN-γ, fresh aliquots of digested lung tissue were stimulated for 5 hours at 37°C, with Cell-stimulation Cocktail Buffer (40.5 μmol/L phorbol 12-myristate 13-acetate [PMA], 670 μmol/L ionomycin, 5.3 mmol/L brefeldin A, and 1 mmol/L monensin; eBioscience), fixed, permeabilized with Cell Permeabilization Buffer (eBioscience) and incubated with anti-mouse IL-12 clone C17.8 (eBioscience) or anti-mouse IFN-γ clone XMG1.2 (Biolegend).

Airways responsiveness

Airway cholinergic responsiveness was assessed by measuring changes in total respiratory system resistance in response to increasing doses of nebulized methacholine. Mice were anesthetized with sodium pentobarbital and a tracheostomy was performed. Mice were mechanically ventilated and total respiratory system resistance was measured plethysmographically with a Buxco FinePointe operating system (Buxco, Wilmington, NC).

Human tracheal aspirate collection and RNA preparation

We examined tracheal aspirates from infants admitted to the C.S. Mott Children's Hospital Newborn Intensive Care Unit, as approved by the University of Michigan Institutional Review Board. Entry criteria included gestational age at birth ≤ 32 weeks, mechanical ventilation for respiratory distress, and age ≤ 7 days. Aspirates were collected during routine tracheal suctioning of mechanically ventilated premature infants as described (24). A 200 μl aliquot of the sample was stored in RLT buffer (Qiagen, Valencia, CA) at −80°C. Total RNA was extracted using the RNeasy Plus Mini kit (Qiagen).

Statistical analysis

Data are represented as mean±SEM or median and interquartile range (IQR). Statistical significance was determined by unpaired t test, nonparametric rank-sum test, one-way or two-way ANOVA, as appropriate. Differences were pointed by Newman-Keuls' multiple comparisons test.

RESULTS

Hyperoxic exposure increases lung inflammatory cytokine and chemokine expression in response to poly(I:C) and RV

Two- to three-day old mice were exposed to hyperoxia or room air for 14 days. Upon discontinuing hyperoxic exposure the mice were treated with PBS, poly(I:C) or RV1B intranasally. Poly(I:C) is a synthetic analog of double-stranded RNA which is recognized by Toll-like receptor (TLR)-3 and mimics respiratory viral infection. Cytokine mRNA (Figure 1A) and protein responses (Figure 1B) were measured five days after treatment. Of the panel of inflammatory cytokines and chemokines examined, only IL-12p40 expression was increased in the PBS-treated hyperoxia-exposed mice. Poly(I:C) treatment of hyperoxia-exposed mice increased mRNA and protein expression of the type 1 cytokines IL-12p40, IFNγ, TNFα, and chemokines CCL2, CCL3 and CCL4, but had no effect on cytokine mRNA expression in air-exposed mice, except for induction of IL-12p40. The increase in IL-12p40 mRNA expression in hyperoxia-exposed, poly(I:C)-treated mouse lungs was significantly greater than that observed after poly(I:C) treatment of air-exposed animals. These results with poly(I:C) suggest the possibility that hyperoxia alters the inflammatory response to viral infection.

Figure 1. Hyperoxic exposure increases lung pro-inflammatory cytokine expression in response to poly(I:C).

Two-to-3 day-old mice were exposed to air or 75% oxygen for 14 days and treated intranasally with 30 μl of intranasal poly(I:C) (1 μg/μl) or PBS immediately after O2 exposure. Five days after treatment control hyperoxia-exposed mice showed increased IL-12p40 mRNA (A) and protein (B) expression. Compared to air-exposed, poly(I:C)-treated mice, hyperoxia-exposed, poly(I:C)-treated mice showed increased mRNA (A) and protein (B) expression of the type 1 cytokines IL-12p40, IFNγ, TNFα, CCL2, CCL3 and CCL4 and increased CXCL1 protein expression (Figure 1B). *P<0.05 versus normoxia, †P<0.05 versus PBS (one-way ANOVA), N=3-5 per group. Results are representative of three independent experiments.

To test this, we examined the lung inflammatory response of hyperoxia-exposed mice to subsequent RV infection. For these experiments, mice were infected with RV immediately following 14 days of hyperoxic exposure and inflammatory cytokine and chemokine mRNA responses were measured 1, 3, 5 and 12 days after infection. As noted above, hyperoxic exposure caused a sustained increase in lung IL-12p40 mRNA expression which lasted up to 12 days after infection (Figure 2A). Similar to the responses to poly(I:C), hyperoxia-exposed mice infected with RV showed significantly higher IL-12p40, IFN-γ, CCL2, CCL3, and CCL4 levels up to 5 days after RV infection. In contrast, five days after RV infection, air-exposed C57BL/6 mice did not show significant increases in cytokine mRNA expression. Cytokine protein levels tended to reflect the mRNA data (Figure 2B and C). Further, compared to air-exposed animals, RV induced mRNA expression of IL-4 and IL-13, type 2 cytokines typically associated with asthma, was decreased after hyperoxic exposure (Figure S1). Together, these results show that hyperoxic exposure amplifies the pro-inflammatory response to RV infection.

Figure 2. Hyperoxic exposure is associated with an exaggerated and sustained pro-inflammatory immune response to RV infection.

Hyperoxic exposure of two to three day old mice is associated with an increase in IL-12 production lasting up to 12 days after exposure (A–C). Compared to air-exposed, RV-infected mice, hyperoxia-exposed mice infected with RV immediately after hyperoxic exposure (A–C) or 7 days after exposure (D) show an exaggerated type 1 immune response with an increased expression of IFNγ, TNFα, CCL2, CCL3 and CCL4. In A, the response to RV was analyzed 1-12 days after infection. In B, the response to RV was analyzed 5 days after infection. In C, whole lung IL-12p70 protein expression was examined 12 days after infection. In D, the response to RV was analyzed 2 days after infection. *P<0.05 versus normoxia, †P<0.05 versus sham infection (one-way ANOVA), N=3-5 per group. Results are representative of three independent experiments.

To further examine the persistence of hyperoxia's effect on RV responses, we infected air- or O2-exposed mice seven days after hyperoxic exposure was discontinued. Compared to air-exposed mice, hyperoxia-exposed mice showed exaggerated cytokine responses to RV infection (Figure 2D).

To define host antiviral responses, we measured mRNAs encoding IFN-α, IFN-β and IFN-λ2. When compared to air exposure, hyperoxia-exposed mice showed equal induction of IFN mRNA levels 1 and 3 days after RV infection. However, IFN mRNA levels were significantly lower in hyperoxia-exposed mice at baseline and 1 and 3 days after RV infection (Figure 3A). There were no differences in IFN levels between air- and hyperoxia-exposed mice 5 days after RV infection.

Figure 3. Hyperoxic exposure is associated with lower IFN mRNA levels in response to RV infection.

A, IFN-α, IFN-β and IFN-λ2 mRNAs were measured by qPCR. When compared to air exposure, hyperoxia-exposed mice showed equal induction of IFN mRNA levels 1 and 3 days after RV infection. However, IFN mRNA levels were significantly lower in hyperoxia-exposed mice at baseline and 1 and 3 days after RV infection. There were no differences in IFN levels between air- and hyperoxia-exposed mice 5 days after RV infection. *P<0.05 versus normoxia-RV infection. B, Viral copy number in the lungs of air- and hyperoxia-exposed mice 1, 3 and 5 days after RV infection was measured by qPCR. Shown are individual data, mean and standard error of the mean. *P<0.001 versus normoxia (unpaired t test), N=4 per group.

We next measured viral copy number in the lungs of air- and hyperoxia-exposed mice 1, 3 and 5 days after infection. The viral copy number was higher in the hyperoxia-exposed mice 1 day after infection, consistent with the lower interferon levels in this group and suggesting potential defective initial antiviral response (Figure 3B). By day 3 and 5 after infection, the viral copy number was similar in both groups, demonstrating that the augmented pro-inflammatory cytokine production in hyperoxia-exposed mice was not due to a sustained increase in viral load (Figure 3B).

Hyperoxic exposure is associated with accumulation of inflammatory cell infiltrates in response to treatment with poly(I:C) or RV infection

To further define the effects of neonatal hyperoxic exposure on the response to subsequent viral infection, we examined mouse lung histology. Compared to control, air-exposed mouse lungs (Figure 4A), the lungs of control, hyperoxia-exposed mice showed fewer and larger alveoli up to 12 days after hyperoxia was discontinued. No other structural abnormalities or lung inflammation were noted on H&E stain. The lungs of hyperoxia-exposed, RV infected mice showed accumulation of small inflammatory cell infiltrates along the distal bronchovascular bundles both 5 and 12 days after RV infection (Figure 4A, black arrows). Lung histological evaluation 5 and 12 days after poly(I:C) treatment also showed small inflammatory cell infiltrates along the distal bronchovascular bundles of hyperoxia-exposed mice (Figure S2). In contrast, the lungs of air-exposed, poly(I:C) or RV infected mice were devoid of such infiltrates.

Figure 4. RV infection of hyperoxia exposed mice leads to increased cholinergic airway responsiveness and accumulation of small inflammatory cell infiltrates along the distal bronchovascular bundles.

A, Representative images of lungs stained with H&E. Original magnification is ×100X. Bar = 100 μm. B, Airway responsiveness 12 days after RV infection of immature mice (N=5–7 per group), *P<0.01 versus hyperoxia control (2-way ANOVA).

RV infection of hyperoxia-exposed mice leads to increase in cholinergic airway responsiveness

Two-to three day old C57BL/6J mice were exposed to air or 75% oxygen for 14 days. On day of life 16, upon discontinuing hyperoxic exposure, mice were infected with RV1B. Airway cholinergic responsiveness was examined on day of life 30 (12 days post infection). Compared to control, air-exposed mice, control, hyperoxia-exposed mice had normal airways responses 12 days after exposure (Figure 4B). In contrast, there was a significant increase in airway cholinergic responsiveness in mice exposed to hyperoxia as neonates and subsequently infected with RV.

Hyperoxic exposure increases the response of specific lung dendritic cell (DCs) populations to RV infection

To define the innate immune cells involved in the exaggerated and sustained cytokine and chemokine response of hyperoxia-exposed mice to RV infection, we examined lungs by flow cytometry, focusing on lung DCs, a potent source of IL-12 (25). Mouse lungs were collected 5 days after RV infection and lung DCs were differentiated from alveolar macrophages (AM) and interstitial macrophages (IM) based on F4/80 and CD11c expression (26). Compared to air-exposure, hyperoxic exposure was associated with increased lung DCs (CD45+, CD11c+, F4/80−) (Figure 5 panel D and G). In addition, we examined the two major subsets of lung CD11c+, F4/80− DCs, CD103+ and CD11b^hi DCs (27). Compared to air-exposure, hyperoxic exposure increased lung CD103+ DCs (Figure 5 panel E and H). This subset of DCs expressed a high level of the activation markers MHCII and CD86 (Figure 5 panel F). Following treatment with RV, the subset of CD103+ DCs was further increased in hyperoxia-exposed mice (Figure 5 panel E and H). The CD11b^hi DCs were not affected by hyperoxic exposure alone, however RV infection of hyperoxia-exposed mice also increased a population of CD11b^hi lung DCs. We investigated whether CD103+ DCs produce IL-12 after RV infection. RV infection after hyperoxic exposure increased the number and percent of IL-12-producing CD103+ DCs (Figure 6, panel A and B). Together, these results indicate that neonatal hyperoxic exposure increases the number of activated, lung-resident CD103+ DCs and this subset is further increased by RV infection. RV infection of hyperoxia-exposed mice was also associated with an increased in the number of CD11b^hi DCs.

Figure 5. Hyperoxic exposure increases the response of specific lung dendritic cell (DCs) populations to RV infection.

Two day-old mice were exposed to hyperoxia for 14 days and subsequently inoculated with RV or sham. Lung digests were subjected to flow cytometry. After gating out debris (A), doublets (B), and nonviable and CD45− cells (C), the viable, CD45+ cell population was examined for lung DCs (F4/80−CD11c+), interstitial macrophages (IMs, F4/80+CD11c−) and alveolar macrophages (AMs, F4/80+CD11c+). (D). Prior hyperoxic exposure increased the percentage of F4/80−CD11c+ cells (G). Hyperoxic exposure increased the percent of CD103+ DCs (E, H). RV infection further increased this subset in hyperoxia-exposed mice. This DC subset showed high activation marker (MHCII and CD86) expression (F). RV infection of hyperoxia-exposed mice also increased the DC CD11b^hi subset (I). *P<0.05 versus control, †P<0.05 versus RV (one-way ANOVA), N=3-4 per group. Results are representative of five independent experiments.

Figure 6. After RV infection, IL-12 producing CD103+ lung DCs increase significantly in hyperoxia-exposed lungs.

Live lung cells were analyzed and gated on CD45+, MHCII^hi, F4/80−, CD11c+) and lung DCs were examined for CD103 and IL-12 expression (A). (B) The percent () of IL-12+CD103+ lung CDs, determined by flow cytometry, is higher with hyperoxic exposure and is further increased after RV infection. *P<0.001 versus normoxia, †P<0.05 versus sham infection (one-way ANOVA), N=3-4 per group. Results are representative of three independent experiments.

Hyperoxic exposure induces lung CD4+ T cell activation and Th1 cytokine production

When derived from CD4+ T cells, IFN-γ is sufficient to mediate Th1 cell development (28). Th1 immune response induces airway hyperresponsiveness in severe asthma (29). Based on the increase in IFN-γ mRNA expression observed in hyperoxia-exposed mice in response to RV, we sought to evaluate the effect of prior hyperoxic exposure on lung CD4+ T cell activation and IFN-γ production in response to RV infection. Using flow cytometry we examined lungs from hyperoxia- or air-exposed mice 5 days after RV infection and quantified CD69, T cell activation marker, and IFN-γ – expressing CD4+ T cells. Hyperoxic exposure increased the population of IFN-γ+CD69+ CD4+ T cells in the lungs (Figure 7). Hyperoxia-exposed mice showed further increase in this population upon RV infection. These results imply that hyperoxic exposure induces lung CD4+ T cells activation and Th1 differentiation.

Figure 7. Hyperoxic exposure induces lung CD4+ T cell activation and IFN-γ production.

Using flow cytometry lungs from hyperoxia- or air-exposed mice 5 days after RV infection were examined. Live lung cells were analyzed and the CD45+, CD3+, TCRβ+, CD4+ T cells were examined for CD69, T cell activation marker and IFN-γ expression; the percent of IFN-γ+ CD69+ lung CD4+ T cells was determined. *P<0.05 versus normoxia, †P<0.001 versus sham infection (one-way ANOVA), n=3-4 per group. Results are representative of three independent experiments.

Expression of Clec9a is increased during hyperoxic exposure

DCs may be activated by DAMPs (for danger-associated molecular patterns). The damaged cell-recognition molecule Clec9a, a C-type lectin, is selectively expressed on CD103+ DCs in both lymphoid and nonlymphoid organs (30, 31). We investigated whether increases in lung CD103+ DCs during hyperoxic exposure are associated with changes in Clec9a expression. We found that lung Clec9a expression was increased at 7 and 14 days of hyperoxic exposure, peaked between day 3 and 5 after the exposure was discontinued and remained increased until 12 days post exposure (Figure 8A). Using qPCR, we examined Clec9a mRNA expression of sorted lung CD103+ DCs and CD11b^hi DCs. CD103+ DCs expressed higher levels of Clec9a (a small amount of Clec9a transcript was identified in CD11b^hi DCs) (Figure 8B). Flow cytometry confirmed Clec9a expression by lung CD103+ DCs, and a higher number of Clec9a+ CD103+ DCs was observed in the lungs of hyperoxia exposed mice (Figure 8C). These data suggest that exposure to high oxygen tension induces expression of Clec9a, by way of increasing the number of Clec9a expressing CD103+ DCs.

Figure 8. Expression of Clec9a is increased during hyperoxic exposure.

A, Lung mRNA expression of the damaged cell-recognition molecule Clec9a, a C-type lectin which is selectively expressed on CD103+ DCs, is increased during and after hyperoxia, *P<.05 (unpaired t test). B, Sorted CD103+ DCs express significantly higher levels of Clec9a mRNA compared to sorted CD11b^hi, P<0.0001, n=3-4 per group (unpaired t test). C, Hyperoxic exposure increases Clec9a+CD103+ lung DCs. D, Pearson correlation analysis was performed using GraphPad Prism 6.0 software and identified that human CLEC9A mRNA expression in tracheal aspirates from premature infants mechanically ventilated for respiratory distress syndrome positively correlated with the degree of prematurity, P<0.01.

Next, we examined CLEC9A mRNA expression in tracheal aspirates from human premature infants mechanically ventilated for respiratory distress syndrome in the first week of life. The degree of prematurity positively correlated with tracheal aspirate CLEC9A mRNA expression (Figure 8D), suggesting that extreme prematurity correlates with expression of the damaged cell-recognition molecule CLEC9A.

Effect of hyperoxic exposure on adenosine metabolism

Damaged cells release ATP and extracellular levels of ATP and its metabolites ADP, AMP and adenosine may each be increased following cellular damage induced by oxidative stress (32, 33). ATP and its metabolites represent another group of DAMP molecules capable of inducing DC activation. Extracellular adenosine acts as a signaling molecule and activates the G protein-coupled adenosine receptors A1 (Adora1), A2a (Adora2A), A2b (Adora2B) and A3 (Adora3) (34). AMP also serves as a ligand for Adora1 (35). Using qPCR, we measured whole lung expression of the four adenosine receptor subtypes, as well as CD73 (ecto-5'-nucleotidase, which converts AMP to adenosine) and CD39 (ectonucleotidase, which hydrolyzes ATP and ADP to AMP). Compared to lungs from air-exposed mice, lungs of hyperoxia-exposed mice showed increased expression of Adora1 and CD73 (Figure 9A) at 7 and 14 days of exposure. After hyperoxia was discontinued, Adora1 expression remained increased. Hyperoxia exposure decreased Adora2a mRNA expression on days 3, 7 and 14 of exposure, increased Adora2b expression at 7 days of exposure and decreased Adora3 at days 3 and 14 of exposure and had no effect on mRNA expression of CD39.

Figure 9. Effect of hyperoxic exposure on adenosine metabolism.

A. Compared to air-exposed mouse lung, lungs of hyperoxia-exposed mice showed increased expression of Adora1 and CD73 at 7 and 14 days of exposure and decreased Adora2a mRNA expression on days 3, 7 and 14 of exposure, *P<.05 (unpaired t test). B,C. Plasma and BALF concentrations of adenosine, AMP, ADP and ATP were determined by mass spectroscopy. After 14 days of hyperoxia serum levels of AMP and AMP:ATP and ADP:ATP ratios in plasma were were increased *P<.05 (unpaired t test).

We next examined plasma and BALF concentrations of adenosine, AMP, ADP and ATP by mass spectroscopy. Following 14 days of hyperoxic exposure, serum levels of AMP were significantly increased (Figure 9B). AMP:ATP and ADP:ATP ratios in plasma were also increased after 14 days of hyperoxic exposure (Figure 9C). A similar tendency was observed in BALF without reaching statistical significance. Together, these results suggest that hyperoxic exposure is associated with changes in ATP release and breakdown and, in particular, AMP/Adora1 signaling. These danger signals could play a role in DC activation and promotion of pro-inflammatory responses.

Finally, we analyzed the effect of hyperoxic exposure on Adora1 mRNA expression in response to RV infection. Adora1 mRNA expression was increased in hyperoxia-exposed mice 5 days after RV infection compared to both sham-infected hyperoxia-exposed mice and air-exposed RV-infected mice (Figure 10). These data indicate that hyperoxic exposure stimulates Adora1 signaling in response to RV infection.

Figure 10. Effect of hyperoxic exposure on Adora1 mRNA expression in response to RV infection.

Adora1 mRNA expression was increased in hyperoxia-exposed mice 5 days after RV infection compared to both sham-infected hyperoxia-exposed mice (*P<0.05, one-way ANOVA) and air-exposed RV-infected mice (†P<0.05, one-way ANOVA).

DISCUSSION

In this report, we show that hyperoxic exposure of neonatal mice causes an exaggerated pro-inflammatory immune response to subsequent RV infection, leading to persistent airway inflammation and hyperresponsiveness. Hyperoxic exposure alone caused sustained increased expression of the inflammatory type 1 cytokine IL-12p40 and transient increased expression of IFN-γ, TNF-α, CCL2, CCL3 and CCL4. Further, infection of hyperoxia-exposed mice with RV led to increased and sustained expression of these cytokines compared to RV-infected, air-exposed mice. Hyperoxic exposure increased the number and activation of lung-resident IL-12-producing CD103+ DCs, which were further increased by RV infection. Hyperoxic exposure increased lung CD4+ T cell activation and IFN-γ production and these were further increased in response to RV infection. Hyperoxic exposure was associated with increased expression of Clec9a, a damaged cell recognition molecule. Clec9a expression was localized to CD103+ DCs and a higher number of these cells was found in the lungs of hyperoxia exposed mice. Finally, hyperoxia increased levels of ATP metabolites and expression of Adora1, further evidence of cell damage and related signaling. These data suggest that hyperoxia causes danger signal-induced DC activation, leading to exaggerated pro-inflammatory airway responses to subsequent viral infection.

IL-12p40 is a subunit of the IL-12 cytokine which is produced by activated DCs and directs Th1 cell development (25). The IL-12 cytokine is a heterodimer composed of two distinct subunits, p40 and p35. When these subunits are assembled together they form biologically active IL-12. In our study, hyperoxic exposure induced overexpression of IL-12p40 mRNA which was further increased by poly(I:C) treatment and RV infection. IL-12p40 and IL-12p70 protein levels were increased in lung and BAL from hyperoxia-exposed mice infected with RV. Hyperoxia-exposed, RV-infected mice also showed increased levels of the type 1 cytokines IFN-γ and TNF-α, as well as increased airway inflammation and hyperresponsiveness. Together these data suggest that, in our model, hyperoxia-induced IL-12 overexpression drives an exaggerated inflammatory response to subsequent respiratory viral infection. Development of a Th1 response and induction of IL-12 or IFN-γ have each been associated with resolution of symptoms in viral infections (36, 37). Although IL-12 appears to be protective during respiratory viral infections, overproduction of IL-12, as observed in sepsis or in the absence of anti-inflammatory IL-10 production, leads to hyperinflammatory responses (38, 39). Furthermore, hyperoxic exposure of mature mice induces IL-12 production and IFN-γ-mediated lung injury (40), consistent with the notion that, under selected conditions, IL-12 could favor an adverse exaggerated type 1 cytokine response to subsequent infections or other adverse stimuli. We believe that, in our model, overproduction of IL-12 leads to similar hyperinflammatory responses.

Next, we endeavored to determine the source of IL-12 production. Mouse lung DCs contain two major subpopulations: CD103+ CD11b^low (or CD11b-negative) DCs and CD103-negative CD11b^high DCs (41–43). CD103+ DCs produce IL-12 and IFN-γ (44–46) and predominantly elicit Th1 and Th17 responses (47), whereas CD11b^high DCs primarily provoke a Th2 response (47, 48). We found that hyperoxic exposure was associated with an increase in lung CD103+, MHCII^hi, CD86^hi DCs, a population of cells which was further increased upon viral infection. CD103+ DCs produced IL-12 when stimulated ex vivo. Together these data suggest that hyperoxia-induced CD103+ DCs are responsible for the exaggerated pro-inflammatory response to subsequent respiratory viral infection. The mechanisms responsible for the increase in lung CD103+ DCs in hyperoxia-exposed mice are not well understood. While it is possible that, under hyperoxic conditions, recruited CD103+ DC progenitors differentiate and become activated (49), we cannot exclude the possibility that the activated CD103+ DCs have an impaired capacity to migrate to the regional lymph nodes as observed during viral infections or in the setting of altered extracellular ATP gradients (50, 51).

Lung DCs form a network of cells that interact with the outside environment and express pathogen-associated molecular pattern (PAMP) and damage-associated molecular pattern (DAMP) receptors that recognize microbes and danger signals. Engagement of PAMP and DAMP receptors triggers DC activation and maturation. Activated DCs upregulate MHC class II and the co-stimulatory molecules CD80 and CD86 and migrate to the regional lymph nodes, where they stimulate naïve CD4+ T cells and direct their differentiation to various T-cell lineages (52, 53). We speculate that hyperoxic exposure generates danger signals which activate CD103+ DCs. Hyperoxic exposure has been associated with lung epithelial cell apoptosis and necrosis (54) and dead cells stimulate DC maturation and migration of mature DCs into draining lymph nodes (55, 56). CD103+ DCs, unlike CD11b+ DCs, express receptors for apoptotic cells along with the machinery to cross-present phagocytosed dead cells (57). Prolonged hyperoxic exposure and resultant oxidative stress are strong inducers of various proinflammatory cytokines in airway cells and pulmonary tissue, and lead to influx of macrophages and neutrophils (58, 59). Here we propose a previously unrecognized mechanism for oxidant-associated inflammatory cell influx: hyperoxia-induced cell injury and death (60) leads to release of “danger signals” that activate the innate immune system and increase inflammatory responses.

Clec9a, also referred to as DNGR-1, is a DC receptor that couples sensing of cell death to innate immune responses. Clec9a diverts dead-cell-associated antigens away from lysosomal compartments to allow retrieval of antigens for cross-presentation to T cells (30, 61, 62). Clec9a appears to recognize particular forms of filamentous actin, perhaps in association with particular actin-binding domains of cytoskeletal proteins (63). Our studies indicate that hyperoxia increases the lung expression of Clec9a, reflecting the increased number of Clec9a+ CD103+ DCs. Clec9a expression is sustained increased after hyperoxia is discontinued, suggesting persistence of a larger population of CD103+ DCs with activated danger signaling. Furthermore, an analysis of tracheal aspirates from premature human infants showed that the level of prematurity positively correlated with respiratory tract Clec9a mRNA expression. Since hyperoxic exposure causes respiratory epithelial cell death (64–66), we speculate that premature infants experience greater epithelial cell death upon exposure to high oxygen tensions relative to that in utero, leading to expansion of Clec9a+ CD103+ DCs.

Following tissue injury, damaged cells release ATP and ADP in the extracellular space, which are then dephosphorylated to AMP by the ectonucleoside triphosphate diphosphohydrolase CD39 (67). AMP is subsequently dephosphorylated to adenosine by the ecto-5'-nucleotidase CD73 (68). Extracellular adenosine has been linked to regulation of inflammation and tissue repair (69). Adenosine signals through one of four G-protein coupled, cell surface receptors (Adora1, Adora2a, Adora2b, Adora3) (70), all of which are expressed on DCs (71), macrophages (72) neutrophils (73) and respiratory epithelial cells (74). AMP also signals through Adora1 (35). Adora1 exerts a pro-inflammatory response, while engagement of Adora2a inhibits adherence and activation of leukocytes (75). Our studies identified increased Adora1 and CD73 expression at 7 and 14 days of hyperoxic exposure, while Adora2a expression was decreased. Adora1 expression remained increased after hyperoxia was discontinued, suggesting persistent activation of danger signaling. Hyperoxic exposure was associated with increased plasma levels of AMP, as well as increased AMP:ATP and ADP:ATP ratios. A similar tendency was observed in BALF. Together these data are consistent with the notion that neonatal hyperoxic exposure induces airway inflammation by increasing ATP metabolism and AMP-Adora1 signaling. Despite increased CD73 expression, adenosine levels were not increased, perhaps due to increased adenosine uptake. Furthermore, we observed significantly increased Adora1 mRNA expression in hyperoxia-exposed mice 5 days after RV infection, suggesting that hyperoxic exposure promotes Adora1 signaling in response to RV infection.

Premature infants and babies with BPD have increased respiratory morbidity including wheezing, asthma, medication use and hospitalizations with respiratory viral infection (1–11). Atopy has not been associated with increased respiratory symptoms in prematurely born infants and infants with BPD (76, 77). The precise mechanisms leading to higher respiratory morbidity and abnormal airway function in prematurely-born children are not well understood. Lung structural abnormalities in the “new,” surfactant-treated BPD are relatively mild, with only hypoalveolarization but no structural airway changes (14–16). These abnormalities do not seem to explain the increased wheezing, airways obstruction, respiratory morbidity or susceptibility to viral infections observed in premature infants. It is conceivable that hypoalveolarization predisposes to airway narrowing by reducing the number of alveolar attachments (78). Airway smooth muscle thickening has been noted in hyperoxia-exposed mice (79, 80) but this finding has not been verified in human infants. We therefore sought an alternative explanation for the observed respiratory morbidity, the details of which we describe above. Few animal studies have investigated the effect of neonatal hyperoxic exposure on response to respiratory viral infection. Short-term neonatal hyperoxic exposure increased the subsequent inflammatory response to influenza A virus infection in mice infected at 8–9 weeks of age (17). Adult mice with prior hyperoxia exposure as neonates showed increased leukocyte recruitment to the lungs and higher levels of CCL2 in BAL fluid up to 9 days after infection. We also found increased CCL2 mRNA expression in hyperoxia-exposed, RV-infected mice. However, we extend these previous results by providing specific cellular and molecular mechanisms responsible for the observed inflammatory response.

We examined the inflammatory response to respiratory viral infection in immature mice, thereby simulating the clinical experience of children with BPD becoming sick with respiratory viruses. Animal studies with neonatal rhinovirus infection demonstrate that depending on age at the time of RV infection, the inflammatory response can be skewed towards Type 2 (< 8 days of age) or Type 1 (> 8 days of age) (23). Our findings are consistent with the above study, as RV infection on day of life 16 induced Type 1 responses.

In our model, the initial priming of the innate immune system by hyperoxia-induced DAMP signaling increased IL-12 expression and directed subsequent hyperinflammatory Th1 responses. Further investigation will be required to determine the requirement and sufficiency of IL-12 for exaggerated airway inflammation and hyperresponsiveness to RV infection. Also, since IL-10 is a potent anti-inflammatory cytokine (81) and negative regulator of IL-12 production (82), future studies examining the capacity of IL-10 treatment to block the observed hyperinflammatory response to RV are warranted.

In summary, early life hyperoxic exposure increased the activation of CD103+ DCs, leading to increased type 1 inflammation and airway hyperresponsiveness upon RV infection. The expression of different DAMP receptors and their ligands was increased during immature mice hyperoxic exposure. In premature infants with a history of BPD, danger signal-induced DC activation may promote pro-inflammatory airway responses, thereby increasing respiratory morbidity.