NF-κB Signaling in Fetal Lung Macrophages Disrupts Airway Morphogenesis Free

During fetal lung morphogenesis, simple epithelial tubes develop into a complex structure competent for gas exchange (1, 2). This process fails to occur normally in preterm infants with bronchopulmonary dysplasia (BPD), a chronic disease that affects ∼60% of preterm infants born before 28 wk (3). BPD results from arrested airway morphogenesis during the canalicular and saccular stages of lung development. Normally during these stages, small terminal airways branch, expand, and divide to form alveolar ducts. In patients with BPD, arrested development leads to fewer saccular stage airways, reduced numbers of alveoli, and lower capacity for gas exchange (4). Infants born after the early saccular stage of development rarely develop BPD, suggesting that the canalicular and early saccular stages represent a window of disease susceptibility. Various environmental factors have been implicated in development of BPD, involving either airway epithelial injury or lung inflammation (5, 6). Understanding how these factors lead to arrested airway morphogenesis will be critical for developing new therapeutic approaches.

Several lines of clinical and experimental evidence suggest infection or injury leads to an inflammatory response that causes or exacerbates BPD. First, chorioamnionitis (infection and inflammation of the amniotic membranes) is detected in up to 70% of preterm deliveries and is associated with increased BPD risk (7). Second, infants that develop BPD often have elevated levels of inflammatory mediators in their airway both at birth and during the later stages of disease progression (8, 9). Third, injecting Escherichia coli LPS into the amniotic fluid of pregnant animals inhibits airway branching and prevents subsequent alveolar development (10, 11). In developing an experimental mouse model for studying the mechanisms leading to BPD, we observed that LPS prevents saccular airway branching in both fetal mice and cultured fetal mouse lung explants. These data suggest innate immune activation and inflammatory signaling intersect with developmental pathways, interfering with processes required for branching morphogenesis.

LPS can inhibit airway branching in fetal lung explants in the absence of circulating inflammatory cells (10), suggesting that resident lung cells are competent to transduce signals that interrupt normal lung development. We have previously identified multiple gene targets of inflammatory signaling that are critical for normal airway morphogenesis (9, 12–14). However, the cellular site of the initial innate immune response critical for disrupting development is not known. Among the cell types in the fetal lung, airway epithelia and vascular endothelia appear at least somewhat capable of responding to microbial products (15, 16). To date, the potential role of macrophages in the fetal lung innate immune response has not been closely examined. Studies involving fetal macrophages have focused primarily on their ability to remove apoptotic cellular debris and remodel extracellular matrix (17–19). In this study, we report that macrophages are the primary cellular sites of the fetal lung innate immune response. Macrophage activation is required for the LPS-mediated production of inflammatory mediators that disrupt airway branching. In addition, targeted activation of fetal lung macrophages, both in cultured lung explants and developing mouse lungs in vivo, inhibits airway morphogenesis and produces a lung phenotype that closely resembles human BPD. Our findings indicate that macrophage activation and subsequent lung inflammation may play a key role in BPD pathogenesis.

Gel-purified E. coli LPS (O55:B5) was purchased from Sigma-Aldrich. The following Abs were used for immunolabeling: rat anti-CD68 (Acris), rat anti-F4/80 (Acris), rat anti–E-cadherin (Zymed), rabbit anti-CD14 (Santa Cruz Biotechnology), and rabbit anti-GFP (Abcam). DAPI, TO-PRO-3 iodide, ProLong Gold antifade reagent with DAPI, and Alexa-conjugated secondary Abs were purchased from Invitrogen. Anti-CD11b Microbeads and magnetic separation equipment and reagents were purchased from Miltenyi Biotec. Reagents for preparing liposomal clodronate were obtained from Sigma-Aldrich. All cell culture media were purchased from Invitrogen. FBS was purchased from Thermo Fisher Scientific.

All animal experiments were thoroughly reviewed and approved by the Vanderbilt University Institutional Animal Care and Use Committee. BALB/cJ and Rosa26-YFP mice were obtained from The Jackson Laboratory. NGL (20), LysM-Cre:IKKβ^F/F (21), and IKFM (L. Connelly, W. Barham, H.M. Onishko, L. Chen, T. Sherrill, T. Zabuawala, M.C. Ostrowski, T.S. Blackwell, and F.E. Yull, submitted for publication) mouse strains were developed and described previously. For timed pregnant matings, embryonal day 0 (E0) was defined as the morning of vaginal plug discovery. To induce transgene expression in IKFM mice, timed pregnant dams were given doxycycline-containing water (2 g/l) ad libitum. E15 fetal mouse lung explants were isolated and cultured as previously described (16). For transgenic models, lungs of each mouse embryo were minced and cultured separately. Approximately 20 explants were isolated from each embryo. All explants studied were between 0.8 and 1.2 mm in diameter at initial time of culture. To quantify saccular airway branching in cultured lung explants, brightfield images of explants were acquired every 24 h of culture. Airway branching was expressed as the number of new branches that formed between 24 and 72 h of culture in each explant. For macrophage depletion using clodronate, control or clondronate-containing liposomes were directly added to explants at the initial time of isolation. LPS was then added to the media 24 h later.

E15 fetal mouse lungs were dissected free of surrounding tissues and placed in ice-cold PBS. The lungs were homogenized and forced through 100-μm and 40-μm cell strainers. Cells were pelleted by centrifugation and resuspended in the presence of anti-CD11b–conjugated Microbeads (Miltenyi Biotec). The CD11b-labeled cells were collected using a magnetic separator, washed, and plated in RPMI 1640 with 10% FBS. Following overnight culture at 37°C in 95% air/5% CO₂, macrophages were washed and treated for 4 h with LPS (250 ng/ml). NGL macrophages were similarly isolated, treated with LPS for various time points, and solubilized in Reporter Lysis Buffer (Promega). Luciferase activity was measured using SteadyGlo reagent (Promega) and a Synergy HT microplate reader (Biotek). Luciferase activity was normalized to total protein content as measured by BCA assay (Promega).

Total RNA isolation, cDNA synthesis, and real-time PCR were performed using standard techniques. Gene expression was compared using the 2^−ΔΔCT method. Independent experiments were performed at least three separate times. Data between groups were compared by ANOVA or Student t test to test for significant differences. IL-1β concentrations were measured in homogenized fetal lung using the Quantikine Mouse IL-1β Immunoassay (R&D Systems). IL-1β concentrations were normalized to total protein as measured by BCA assay (Pierce).

Mouse lung tissue, fetal lung explants, and cultured macrophages were fixed, processed, and immunolabeled using standard techniques. For immunofluorescence, Alexa-conjugated secondary Abs were used for visualization, and nuclei were alternatively labeled with DAPI or TO-PRO-3. Immunohistochemical processing was performed using Vectastain Kits and 3,3′-diaminobenzidine visualization (Vector Laboratories) with Mayer’s H&E counterstaining (Sigma-Aldrich).

Confocal images were acquired using either an Olympus FV1000 (Olympus) or Leica SPE (Leica Microsystems) laser scanning confocal microscope. Widefield fluorescence images and brightfield images of fetal mouse lung explants were obtained using an Olympus IX81 microscope equipped with a Hamamatsu Orca ER CCD monochrome camera and Slidebook software (Olympus). Color images of lung specimens were photographed using a Nikon TE800 and SPOT color CCD camera (Diagnostic Instruments). All microscopy images were saved in the Tagged Image File format and imported into Photoshop (Adobe Systems) for processing. Images for comparison were always identically processed. For live imaging of NF-κB activation, NGL explants were cultured in a stage-top incubator (Okolab) and imaged using an Olympus IX81 inverted microscope (Olympus) with a WeatherStation enclosure. Exact temperature, humidity, and CO₂ concentration were maintained during the imaging experiment. Fluorescence images were acquired every 10 min for the first 2 h and then every 20 min for 2–24 h of culture. Time series images were processed in Slidebook (Olympus). Lung morphometery and cell counting were performed using the Image Processing Tool Kit (Reindeer Graphics) within Photoshop (Adobe Systems). Fractional lung volumes were measured by analyzing images from serial fetal lung sections using a counting grid function. The fractions of airspace, large airway epithelia, small or distal airway epithelia, and mesenchyme were measured on E-cadherin–labeled sections. Septal thickness in newborn lungs was measured using the global intercept function. To account for possible anisotropy, intercepts in each image were measured over multiple iterations with 10° of rotation between each measurement. The mean intercept length for each image was recorded.

Where LPS initially activates the innate immune system in fetal lungs has not been identified. We therefore investigated the localization and kinetics of LPS-induced NF-κB activation in E15 fetal lung explants from NF-κB transgenic reporter mouse (NGL), which expresses GFP and luciferase downstream of a promoter synthesized from the NF-κB–binding repeats within the HIV-1 long terminal repeat (20). Time-lapse imaging of NGL explants demonstrated NF-κB–dependent GFP expression by 90 min after LPS treatment, with continued GFP expression to 12 h (Fig. 1A–D, Supplemental Movies 1–4). GFP expression localized to cells throughout the lung mesenchyme. To visualize the cellular site of NF-κB–GFP expression at higher resolution, we imaged NGL explants using confocal microscopy. As seen in Fig. 1E–M, 2 h of LPS treatment stimulated GFP expression predominantly in cells that colabeled with the macrophage marker CD68. Following 12 h of LPS treatment, GFP reporter expression was detected both in CD68-positive macrophages and in adjacent mesenchymal cells (arrows in Fig. 1G, 1M). The close proximity of GFP-expressing mesenchymal cells to CD68-postive macrophages suggests release of secondary inflammatory mediators that could then activate NF-κB in nearby cells.

Because macrophages were present in E15 fetal lungs and positioned to regulate inflammatory responses, we examined the timing of macrophage appearance into the fetal mouse lung during development. Immunostaining identified CD68-positive cells in the lung mesenchyme as early as E10 during the initial steps of lung formation (Fig. 2A). Macrophages remained present in the mesenchyme throughout fetal development. Fetal lung macrophages also expressed F4/80 (Fig. 2B). To measure function, we tested if LPS could directly activate isolated fetal lung macrophages. We obtained macrophages from E15 lungs using anti-CD11b–coated magnetic beads (Fig. 2C), exposed them to E. coli LPS, and measured gene expression by RT-PCR. LPS stimulated macrophage expression of the inflammatory mediators IL-1β and TNF-α, but did not affect expression of TGF-β or BMP4 (Fig. 2D). Using macrophages isolated from E15 NGL mouse lungs, we next showed that LPS directly activates NF-κB–dependent luciferase activity (Fig. 2E). These data clearly demonstrate that macrophages within the fetal mouse lung respond to LPS and can produce inflammatory mediators.

We have previously developed an E15 fetal mouse lung explant model to study the mechanisms regulating canalicular and saccular stage airway branching (10, 16). Following 72 h of culture, the structures along the periphery of these explants resemble early saccular lungs by undergoing type I/type II cell differentiation, thinning of the mesenchyme, and branching of small airways. This model parallels the period of lung development occurring postnatally in extremely preterm infants born between 23 and 28 wk gestation, making it relevant for investigating BPD pathogenesis. Airway branching in E15 explants requires fibronectin, mesenchymal α₈β₁-integrin expression, and fibroblast growth factor-10 (FGF-10) (10, 12, 13). We have also previously shown that TLR agonists, hyperoxia exposure, and tracheal aspirate fluid from preterm infants exposed to antenatal inflammation can each inhibit airway branching in cultured explants (10, 13, 22). To initially test if the inhibitory effects of LPS on branching required macrophage activation, we depleted macrophages from E15 fetal lung explants using clodronate-containing liposomes (Fig. 3). We then measured the number of new airways that formed along the periphery of each explant between 24 h and 72 h. Although LPS inhibited formation of new saccular airways in control and empty liposome-treated explants (Fig. 3E, 3G), clodronate depletion of macrophages protected saccular airway branching from the effects of LPS (Fig. 3F, 3G). Macrophage depletion also substantially reduced the amount of LPS-induced IL-1β within lung explants (Fig. 3H), suggesting macrophages produce most of the IL-1β within LPS-treated fetal lung explants. Macrophages, therefore, are important for both production of inflammatory mediators within the fetal lung and inhibiting airway branching following exposure to LPS.

As clodronate could have off-target toxic effects, we next used a molecular approach to test if NF-κB activation in macrophages was required for LPS-dependent disruption of saccular airway branching. For these experiments, we used mice with macrophage-specific IκB kinase β (IKKβ) deletion. LysM-Cre:IKKβ^f/f mice (IKKβΔ) lack functional IKKβ in myeloid cells, including macrophages and monocytes (21). LysM-Cre recombinase activity was confirmed in fetal lung explant macrophages using Rosa26-YFP reporter mice (Fig. 4A, 4B). We next measured the effects of LPS on explants from IKKβΔ and littermate controls. LPS failed to inhibit saccular airway branching in IKKβΔ explants (Fig. 4C, 4D), supporting the critical role of macrophages in inflammation-dependent disruption of airway morphogenesis and identifying NF-κB as the operative signaling pathway in mediating this effect. The total amount of IL-1β produced in LPS-treated IKKβΔ explants was much lower than in explants from littermate controls, again demonstrating that macrophages are responsible for a majority of the IL-1β produced in the LPS-treated fetal lung (Fig. 4E).

Having observed that activation of the NF-κB pathway in macrophages appears to mediate the effects of LPS on saccular airway morphogenesis, we next tested if NF-κB activation in macrophages is sufficient to disrupt airway morphogenesis and whether macrophages can impact airway formation in vivo. For these experiments, we used IKFM transgenic mice that express a macrophage-specific doxycycline-inducible c-fms transactivator and a constitutively active IKKβ mutant (cIKKβ) gene downstream of a tet-responsive promoter (L. Connelly et al., submitted for publication). We cultured saccular stage fetal lung explants from IKFM mice and control littermates in the absence or presence of doxycycline (Fig. 5A, 5B). Doxycycline had no effect on branching in control explants, but caused airway dilation and reduced branch formation in IKFM explants (Fig. 5B). Excessive IKKβ activity in IKFM macrophages inhibited branching, as did adding LPS (Fig. 5C). Inducing cIKKβ expression in macrophages increased IL-1β expression in fetal lung explants, demonstrating activity of the transgene (Fig. 5D). These results suggest that NF-κB activation specifically in macrophages is sufficient to disrupt normal fetal lung morphogenesis.

We next tested if macrophage-specific overexpression of cIKKβ in IKFM mice could alter lung morphogenesis in intact fetal mice. Pregnant IKFM mice were treated with doxycycline to activate macrophages during either the late pseudoglandular stage (E12–E16) or from the late pseudoglandular stage through the canalicular and early saccular stages (E14–E18, Fig. 5E). Doxycycline administration from E12–E16 did not cause apparent changes in airway formation (Fig. 5E, 5F). However, transgene activation from E14–E18 caused significant changes with less airspace and increased mesenchyme in IKFM lungs compared with littermate controls (Fig. 5E, 5G). These differences were consistent with defects in saccular airway branching morphogenesis. The lack of effect in mice treated E12–E16 suggests that airway branching in the E14–E18 lung may have a window of vulnerability to macrophage activation. These data support the idea that macrophage activation and subsequent lung inflammation disrupt the later stages of airway formation.

To investigate how the morphological changes observed in IKFM mice affect respiratory function, mice treated with doxycycline from E14–E18 were allowed to deliver at E19 and E20. Disruption of normal lung development in IKFM pups led to reduced survival in the first 24 h. Newborn IKFM mice appeared cyanotic, and many died soon after birth, whereas control littermates appeared healthy (Fig. 6A, 6B). Histological examination of newborn IKFM lungs again showed a thickened lung interstitium compared with control littermates (Fig. 6C, 6D). As with mice harvested at E18, the microscopic appearance of newborn IKFM mouse lungs, with thickened interstitium and simplified airways, showed similarities to the lungs of patients with BPD. Macrophages were present both lining the airway lumen and within the interstitium in both control and IKFM newborn lungs (Fig. 6E). IKFM mice had increased numbers of F4/80-positive cells, which may represent increased macrophage recruitment, maturation, or proliferation (Fig. 6F). IKFM and control lungs contained similar numbers of lymphocytes and neutrophils (Fig. 6G), without signs of consolidated infiltrates or edema. These data further demonstrate that macrophage NF-κB activation, which could occur secondary to infection or other insults, can inhibit saccular airway morphogenesis.

Macrophage activation from E14–E18 in IKFM mice caused a persistent inhibition of expression of multiple genes critical for branching morphogenesis in newborn lungs (12, 23, 24), including reduced expression of the α₈ integrin subunit, BMP4, and Wnt7b (Fig. 6H). Expression of BMPR1A and FGF-10 also trended lower in IKFM lungs. Macrophage activation and release of inflammatory cytokines can therefore inhibit gene expression in the developing lung, leading to altered airway morphogenesis. These findings are consistent with our findings in lung explants, where macrophage activation can inhibit airway branching in the absence of systemic circulation. Activation of NF-κB in fetal lung macrophages could therefore be a common feature of insults that disrupt normal airway morphogenesis. As such, our model of a BPD-like phenotype in IKFM mice may be useful in further defining the pathogenesis and course of BPD.

In this study, we demonstrate that NF-κB activation in macrophages is a key initial step in the fetal lung inflammatory response and sufficient to disrupt fetal lung morphogenesis. Macrophages in the canalicular or saccular stage fetal lung reside in the interstitium and respond to endotoxin. The NF-κB–dependent release of inflammatory mediators from these activated macrophages may then cause a second wave of signaling in neighboring cells, including the mesenchymal cells critical for airway morphogenesis. Even when the entire fetal lung is exposed to LPS, as is the case when explants are treated with LPS, macrophages appear to be the major cellular site of NF-κB activation and cytokine production. Macrophage depletion or targeted IKKβ deletion dramatically reduced the amount of IL-1β produced in response to LPS treatment and protected airway branching from the inhibitory effects of LPS. In addition, targeted activation of fetal lung macrophages was sufficient to disrupt lung morphogenesis, causing perinatal lethality and airway pathology that resembles human BPD.

These findings fill an important gap in our knowledge of how the immature fetal lung responds to inflammatory stimuli. When preterm infants are exposed to chorioamnionitis in utero, inflammatory cytokines accumulate in their airways (9). The infants with high levels of inflammatory mediators in their lungs at birth are more likely to develop BPD (25). Not only are macrophages the major source of inflammatory mediators in the lung, but also their location in the fetal lung interstitium may increase the likelihood that macrophage-derived cytokines can affect immediately adjacent cells. Recent studies have demonstrated increased macrophage recruitment to fetal or newborn lungs following either mechanical or endotoxin-mediated injury (26, 27). However, our current study is the first to connect macrophage activation and altered lung morphogenesis, to our knowledge. We propose a two-wave mechanism for NF-κB activation leading to arrested lung development in BPD. In this model, microbial products initially activate NF-κB in lung macrophages. The release of inflammatory mediators, particularly IL-1β and/or TNF-α, then causes NF-κB activation in the adjacent mesenchymal cells. NF-κB activation in the mesenchyme disrupts expression of genes important for the precise, controlled epithelial–mesenchymal interactions that regulate airway branching. This mechanism is supported by our previous findings that NF-κB activation in fetal lung mesenchymal cells interferes with normal expression of important developmental genes, including FGF-10 and integrin α₈β₁ (9, 12–14). The decreased BMP4 expression in IKFM lungs may be due to the direct effect of inflammatory cytokines on the airway epithelium (28) or secondary to reduced FGF-10 expression in the mesenchyme (23). Although inflammatory mediators that signal via other pathways may also affect lung development, those that can activate NF-κB in target cells appear to be most detrimental.

The IFKM mouse strain may provide an important model for further studying how macrophage activation and inflammation arrests lung development, particularly if the phenotype can be modulated so that affected pups can survive the neonatal period. The inducible cIKKβ transgene allows macrophage activation at distinct stages of lung development, as compared with postnatal rodent models that are restricted to studying the late saccular and alveolar stages of lung development (29). Human infants born at a comparable stage of lung development to that of newborn mice very rarely develop BPD, making these experimental models less useful for investigating the early mechanisms of BPD pathogenesis (3). IKFM mice may also allow us to test if the effects of macrophage activation are reversible or persistent, how repetitive activations affect lung development, and if macrophage activation alters the susceptibility to other types of injury including hyperoxia, mechanical trauma, and viral infection. These future studies may help us better understand why some infants develop only mild BPD and others have more severe or progressive disease. The IKFM experimental model will potentially permit studies that target different phases of the inflammatory response in the fetal and newborn lung to better identify potential strategies for treatment.

Although fetal lung macrophages clearly respond to innate immune stimuli, their functional role in normal embryogenesis remains uncertain. In developing tissues, macrophages may remove cellular debris, remodel extracellular matrix, or express growth factors (30). Removal of apoptotic cells may be critical for lung morphogenesis, as at least one study has demonstrated that mice lacking the phosphoserine receptor are unable to phagocytose apoptotic cells and die following birth with abnormal lung structure (31). In the developing mammary gland, macrophages localize to collagen fibrils along elongating ductal branches, possibly playing a role in extracellular matrix remodeling (32). Macrophages present in the fetal kidney may contribute to development by expressing specific Wnt ligands (33). We do not yet know if fetal lung macrophages play a similar trophic role in normal airway morphogenesis. In addition to the downstream effects of macrophage-derived cytokines on mesenchymal cell gene expression, inflammatory activation could divert macrophages from their yet uncharacterized developmental or trophic roles. This intersection of inflammatory and developmental pathways could therefore be a focal point for disease pathogenesis.

The two-wave model for propagation of inflammatory signals that disrupt branching morphogenesis suggests several potential therapeutic strategies for preventing BPD. Targeting macrophage activation could prevent the initial wave of fetal lung inflammation. The feasibility of this approach is supported by our experimental data using macrophage depletion and targeted deletion of IKKβ. However, many women clinically present already in preterm labor and with evidence of chorioamnionitis (34). In these situations, fetal lung macrophage activation may have occurred prior to delivery. Therefore, alternative strategies could target inflammatory mediators such as IL-1β and TNF-α to prevent the effects of inflammation on mesenchymal gene expression and altered epithelial–mesenchymal interactions. Finally, approaches that restore the mesenchymal cell phenotype to a normal fetal lung gene expression pattern and cell behavior could promote ongoing lung morphogenesis even in preterm infants with previous inflammatory exposures.

We thank Dr. Michael Karin for the LysM-Cre:IKKβ^F/F mice. We also thank Riet van der Meer, Li Zhang, Amanda Stinnett, Kassidy Thibodeau, and Lianyi Chen for assistance and scientific colleagues for helpful advice and comments.

The authors have no financial conflicts of interest.