Brief mechanical ventilation causes differential epithelial repair along the airways of fetal, preterm lambs

Abstract

Mechanical ventilation of preterm lambs causes lung inflammation and injury to the airway epithelium, which is repaired by 15 days after ventilation. In mice, activated basal cells (p63+, KRT14+, KRT8+) initiate injury repair to the trachea, whereas club cells coordinate distal airway repair. In both human and sheep, basal cells line the pseudostratified airways to the distal bronchioles with club cells only present in terminal bronchioles. Mechanical ventilation causes airway epithelial injury that is repaired through basal cell activation in the fetal lung. Ewes at 123 ± 1 day gestational age had the head and chest of the fetus exteriorized and tracheostomy placed. With placental circulation intact, fetal lambs were mechanically ventilated with up to 15 ml/kg for 15 min with 95% N₂/5% CO₂. Fetal lambs were returned to the uterus for up to 24 h. The trachea, left mainstem bronchi, and peripheral lung were evaluated for epithelial injury and cellular response consistent with repair. Peripheral lung tissue had inflammation, pro-inflammatory cytokine production, epithelial growth factor receptor ligand upregulation, increased p63 expression, and proliferation of pro-SPB, TTF-1 positive club cells. In bronchi, KRT14 and KRT8 mRNA increased without increases in Notch pathway mRNA or proliferation. In trachea, mRNA increased for Notch ligands, SAM pointed domain-containing Ets transcription factor and mucin 5B, but not for basal cell markers. A brief period of mechanical ventilation causes differential epithelial activation between trachea, bronchi, and peripheral lung. The repair mechanisms identified in adult mice occur at different levels of airway branching in fetal sheep with basal and club cell activation.

very preterm infants often require mechanical ventilation during the canalicular and saccular stages of lung development, and this ventilation contributes to both bronchopulmonary dysplasia (BPD) and airway changes that can persistent into at least adolescence (10, 23, 25). Although neonatologist have begun using less invasive respiratory support in the delivery room, up to 12% of preterm infants still receive large tidal volume (>10 ml/kg) ventilation during resuscitation (40, 45). Since most infants with BPD eventually come off oxygen therapy and alveolar numbers will increase, the long-term changes in the airways of survivors of BPD may cause more morbidity than the initial injury to the distal lung tissue (14, 23). School-age children with a history of BPD have decreased forced expiratory volume in 1 s, increased respiratory symptoms, and decreased peak flow measurements (14). Mechanical ventilation at birth in preterm sheep stretches the airways and causes airway epithelial injury and diffuse lung inflammation (16, 18, 21). Prolonged mechanical ventilation of preterm fetal sheep alters the lung structure similarly to the histopathology seen in infants with BPD (41). If the injurious mechanical ventilation is brief, preterm fetal sheep can repair the airways by 15 days after the insult (4, 5). Abnormal cellular and airway remodeling may be the initial repair process that progresses to lifelong airway abnormalities in survivors of BPD. Since lung injury may be difficult to avoid in preterm infants, understanding the pathophysiology and mechanisms of repair is essential for the development of effective therapies to decrease the morbidity and economic burden of BPD.

Elegant studies using transgenic mice and various injury models have demonstrated that airway epithelial injury is repaired through a coordinated series of cellular proliferations and trans-differentiations, in specific cell types in different regions of the airways (22, 29, 43). Although most of the cell types described in rodents are present in humans, the distribution of these cells differs significantly along the airways (39). Basal cells compose about 30% of the pseudostratified mucociliary epithelium of the lung and can act as progenitor cells for most of the cell types in the epithelium (22, 43). The epithelium in the mouse trachea is repaired through activation of the p63+, keratin 5+ (KRT5) basal cells into KRT14+, KRT8+ basal cells (43). The distribution and cell types of the human and sheep bronchi down to roughly the fourth generation bronchi are similar in composition to the mouse trachea, with submucosal glands in the cartilaginous airways (13). However, in the mouse, few basal cells are present in the distal lungs, and airway repair relies on dedifferentiation of club cells into more primitive cells before proliferation and transdifferentiation (22). In contrast, the human and sheep lungs have very few club cells in more distal bronchioles, whereas basal cells continue to exist (8, 13). By the terminal bronchioles in humans, there are some club cells containing club cell secretory protein (CCSP) and prosurfactant B. The epithelial lining of preterm airways contains most of the cell types found in the adult, but the repair mechanisms from mechanical ventilation injury have not been evaluated for the preterm lung (5).

To identify the molecular pathways involved in airway repair in the preterm, we caused airway injury with mechanical ventilation in preterm fetal sheep and assessed recovery from injury (16, 18, 21). The fetal model system maintained placental blood supply and allowed return of the fetus to the uterus for evaluation of the progression of injury and repair independent from additional injury from continued mechanical ventilation or oxygen exposure following preterm delivery (17, 18, 21). We tested the hypothesis that airway epithelial injury will be repaired in the preterm animal through activation of the basal cells throughout the airways. We evaluated multiple regions of the airways over several time points after injury to evaluate the inflammatory and basal cell responses.

METHODS

Maternal anesthesia and fetal thorax exposure.

All animal experiments were performed with approval of the Animal Ethics Committees of the University of Western Australia. Date-mated Merino Ewes at 123 ± 1 days gestational age (GA; term is ∼150 days GA) were premedicated with xylazine (0.5 mg/kg im), ketamine (5 mg/kg iv), and midazolam (0.25 mg/kg iv). Ewes were intubated, and anesthesia was maintained with isofluorane (0.5–2% in 100% O₂), which crosses the placenta and anesthetizes the fetus (17). A midline hysterotomy was performed to allow exteriorization of the fetal head and chest up to the umbilical cord (18). An endotracheal tube was tightly secured through a tracheostomy and free-flowing fetal lung fluid was gently removed with an 8 F catheter (average 32 ml). Placental circulation remained intact throughout the ventilation and recovery of the ewe.

Fetal interventions.

The fetal lambs were randomly assigned to either mechanical ventilation for 15 min or control (PEEP only). Each animal was positioned prone on the abdomen of the ewe to maintain placental circulation. The fetal lambs were ventilated with a Fabian ventilator (Acutronic, Switzerland) with initial settings of a rate of 50, inspiratory time of 0.7 sec, PEEP of 0 cmH₂O, and an initial peak inspiratory pressure of 40 cmH₂O. The weight of the animal was estimated, and peak inspiratory pressure was adjusted (max 55 cmH₂O) to achieve a target of 7 ml/kg tidal volume by 4 min of ventilation, 12 ml/kg by 8 min, and 15 ml/kg from 10 to 15 min (16, 18). Ventilator variables were chosen to cause airway and peripheral lung injury (18). Fetal lambs received heated and humidified 95% N₂ + 5% CO₂ to avoid oxygen exposure and changes in fetal Pco₂. After the 15-min intervention, the tracheostomy was removed and the fetus was returned to the uterus for: 30 min, 2 h, 5 h, 10 h, or 24 h (n = 4/group). In animals receiving 30-min or 2-h recovery, the ewe remained under general anesthesia. In fetal lambs assigned to longer postinjury periods, the ewes were recovered from anesthesia and placed back in cages with fentanyl analgesia for pain control. Control lambs received the fetal surgery, tracheostomy, a PEEP 2 cmH₂O but no tidal volume ventilation, and return to uterus for 30 min, 10 h, or 24 h (n = 2/time point). Controls were pooled for comparisons since no differences were seen between groups. At the assigned time interval, the ewe and fetus were euthanized with pentobarbital (100 mg/kg), and fetal samples were collected for molecular analysis. Cord blood gases were measured with a Siemens Rapidlab 1265 (Siemens, Australia).

Lung processing and BAL analysis.

At autopsy, a deflation pressure-volume curve was measured from a pressure of 40 cmH₂O (24). Bronchioalveolar lavage fluid (BALF) of the left lung was collected by repetitive saline lavage (32), and used for cell counts and differential analysis. Cytospins were stained with Diff Quick (Fischer Scientific) for differential cell counts on 200 cells/slide (24). Tissues from the right lower peripheral lung, left mainstem bronchi with surrounding lung parenchyma removed, and trachea were snap frozen for RNA isolation (28). The right upper lobe was inflation fixed at 30 cmH₂O with 10% formalin and then paraffin embedded (28). A portion of the trachea and right mainstem bronchi were formalin fixed prior to paraffin embedding.

Quantitative RT-PCR.

Messenger RNA was extracted from the trachea, the left mainstem bronchus, and peripheral lung tissue from the right lower lobe with TRIzol (Invitrogen). cDNA was produced from 1 μg mRNA by using Verso cDNA kit (Thermoscientific). We used custom Taqman gene primers (Life Technologies) for ovine sequences for amphiregulin (AREG), epiregulin (EREG), heparin-Binding EGF (HB-EGF), Interleukin 1β (IL-1β), IL-6, IL-8, Hes1, Hey1, Hey2, keratin 5 (KRT5), KRT8, KRT14, mucin 5B (MUC5B), p63, SAM pointed domain-containing Ets transcription factor (SPDEF), and CCSP (15, 19, 20). Quantitative RT-PCR was performed with iTaq Universal mix (Bio-Rad) in a 15-μl reaction on a CFX Connect machine and software (Bio-Rad). 18S primers (Life Technologies) were used for the internal loading control. Results are reported as fold increase over mean for the control animals. Since the number of basal cells likely varies based on sampling, mRNA results for basal cells markers (KRT14, KRT8, KRT5) were normalized to both p63 and KRT5 values for some analysis and reported as ratios.

Western blot analysis.

Protein concentrations from lung tissue were determined by Bio-Rad Protein Assays (7). Forty micrograms of protein were denatured in BME at 95° for 5 min, then run on Tris-glycine 10% gel and transferred to a 0.45-μm nitrocellulose membrane (Bio-Rad). Membranes were blocked with 5% normal milk fat and then incubated with p63 1:500 (Santa Cruz) or β-Actin 1:2,000 (Thermo Scientific) overnight at 4°. Appropriate IgG-HRP secondary antibodies were added at 1:10,000 dilutions. Membranes were developed with enhanced chemiluminescence (Pierce/Thermo Fisher Scientific) and then imaged on Syngene PXi multigel imaging system (Syngene) and quantified with ImageJ 1.48v (National Institutes of Health). p63 And β-actin were analyzed on the same gels without membrane stripping, and reported as p63/β-actin with results reported as fold increase over control.

Immunohistochemistry.

Immunostaining protocols used paraffin sections (4 μm) of formalin-fixed tissues that were antigen retrieved in heated citrate and pretreated with 3% hydrogen peroxide (26, 27). The sections were incubated with rabbit antihuman Ki67 1:250 dilution (Thermoscientific), goat antihuman p63 1:500 (R&D Systems), mouse antihuman pro-SPB 1:100 (Seven Hills), or rabbit antihuman TTF-1 1:200 (Seven Hills) in 4% horse serum overnight. For immunohistochemistry using biotin labeled secondary antibody, immunostaining was visualized by Vectastain ABC Peroxidase Elite kit to detect the antigen:antibody complexes (Vector Laboratories). The antigen detection was enhanced with nickel-DAB, followed by TRIS-cobalt and the nuclei counterstained with nuclear fast red (27). Fluorescent double staining was performed overnight at 4° with p63 1:100 dilution, KRT5 1:100 dilution, then rabbit antigoat fluorescent antibody Alexa 594 1:200 (Thermoscientific) for 1 h, followed by blocking in 5% horse serum for 1 h, then goat antihuman Ki67 1:100 followed by 1 h with secondary antibody Alexa 488 1:200 (Thermoscientific). Slides were DAPI counterstained and visualized on a Leica DM 4008b microscope. TUNEL assay for assessment of apoptosis was performed on right upper lobe segments per manufactures protocol (TACS-XL kit, R&D Systems). Positive control for TUNEL were created using TAC Nuclease (R&D Systems). Slides for Ki67 and TUNEL were blinded and 10 segments analyzed based on 0) absent, 1) slight, 2) moderate, or 3) large number of positive cells in the peripheral lung, and epithelial cells from terminal airways were counted and reported as percent of % cells.

Data analysis and statistics.

Results are shown as means ± SE. Statistics were analyzed with Prism 6 (GraphPad) by using Student's t-test, Mann-Whitney nonparametric, or ANOVA tests as appropriate. Significance was accepted as P < 0.05.

RESULTS

All fetal lambs survived the fetal ventilation, return to the uterus, and the interval to delivery. There were no differences in birth weights, gender, or GA (123 ± 1 days) between groups. All ventilated animals achieved similar escalating tidal volumes during the 15-min procedure (Table 1). Control lambs received only a PEEP of 2 cmH₂O and no tidal volume. The peak pressures needed to achieve large tidal volumes were the maximum value of 55 cmH₂O in many animals and thus the V_T/kg at 15 min were somewhat lower than the target of 15 ml/kg (Table 1). Since the entire fetal chest was exteriorized, lung expansion was limited by low compliance of the lungs and not chest compression. Low compliance would be expected in preterm lambs without antenatal steroids or surfactant treatment and correlates with the low volumes at 40 cmH₂O (V40) on postmortem pressure-volume curves (Table 1). The lower V40 values at 10 and 24 h after intervention are consistent with lung injury and inflammation.

Table 1.

Characteristics and ventilation variables

			VT/kg, ml			PIP
Group	N	BW, kg	5 min	10 min	15 min	15 min	V40, ml/kg
Controls	6	3.19					8.0
30 min	4	3.34	7.1	11.9	14.7	50.8	14.6
2 h	4	3.19	6.7	10.6	12.4	54.8	6.1
5 h	4	3.02	6.7	10.4	11.9	54.3	4.3
10 h	4	3.05	7.1	11.8	13.6	54.0	3.3*
24 h	4	3.43	7.4	12.0	14.3	51.8	3.2*

Values are given as means per group. BW, birth weight; V_T, tidal volume; PIP, peak inspiratory pressure; V40, volume at 40 cmH₂O.

^*P < 0.05 vs. controls.

Lung inflammation and EGFR ligand activation.

Pro-inflammatory cytokine mRNA's for IL-1β, IL-6, and IL-8 in the peripheral lung tissue were increased 30 min after the injury with mechanical ventilation (Fig. 1, A–C). The mRNA levels for the cytokines were highest at 2 h after ventilation injury and slowly decreased toward baseline levels by 24 h after the intervention (Fig. 1, A–C). IL-1β mRNA levels remain elevated even 24 h after intervention (Fig. 1A). In response, inflammatory cells were present in the BAL fluid collected at autopsy by 30 min in ventilated animals but not in controls. Inflammatory cells were predominantly neutrophils (54% of cells in BAL) at 2 h, but by 10 h the number of monocytes and macrophages increased (83% of cells), and predominantly monocytes and macrophages (88%) were present by 24 h. The mRNA levels for ligands of the epidermal growth factor receptor (Fig. 1, D–F) amphiregulin (AREG, Fig. 1D), epiregulin (EREG, Fig. 1E), and heparin-binding epidermal growth factor (HB-EGF, Fig. 1F) mRNA increased with mechanical ventilation in the peripheral lung tissue.

Fig. 1.

Pro-inflammatory cytokine and EGFR ligand mRNA in peripheral lung tissue. A–C: pro-inflammatory cytokine mRNA for IL-1β, IL-6, and IL-8 increase by 30 min with highest levels at 2 h. Epidermal growth factor receptor ligands mRNA increase with mechanical ventilation. D: amphiregulin (AREG) mRNA increased by 30 min and remains elevated to after 10 h. E: epiregulin (EREG) mRNA increased in peripheral lung but had more variability between animals. F: heparin-binding epidermal growth factor (HB-EGF) mRNA has a smaller increase than AREG and EREG and returns to baseline by 5 h. *P < 0.05 vs. control animals.

Airway epithelial cell proliferation and apoptosis.

There was increased airway congestion and parenchymal edema in the right upper lobe as the time interval from injury increased (Fig. 2), which is consistent with previous studies on lung inflammation. Apoptosis, as indicated by TUNEL assay, increased in the peripheral lung tissue from 2 to 24 h (Fig. 2, A, D, and E). The cells of the bronchioles were negative for TUNEL staining (Fig. 2, B, D, and E), demonstrating these cells were not undergoing apoptosis. Ki67 protein, a marker of cellular proliferation, increased in the bronchiole cells and in cells within the parenchyma of the lung (Fig. 2, F–J). By 24 h 56% of the epithelial cells lining bronchioles were Ki67 positive (Figs. 2J and 3E). CCSP mRNA decreased in the peripheral lung fourfold in the 24-h animals relative to controls animals (Fig. 3A). The proliferation at 24 h in the terminal bronchioles may be due to club cell proliferation and differentiation, as these cells initially costain with pro-SPB and TTF-1 (Fig. 3, B and C). The protein found in all basal cells, p63, increased in the peripheral lung tissue (Fig. 4A) as early as 30 min after ventilation and remained elevated to 24 h. Basal cells in the medium-sized airways were labeled with p63 (Fig. 4, C and F, red cells) and KRT5 (Fig. 4I), but do not costain with Ki67 (Fig. 4E, green cells), indicating that the basal cells were not proliferating. Neither p63+ nor KRT5+ cells were clearly identified in the distal airspaces at any time point (Fig. 4).

Fig. 2.

Apoptosis and proliferation in peripheral lung. A–E: terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining increases in the lung parenchyma (A) by 5 h (D) compared with control animals (C), and remains elevated at 24 h (E). There was no increased TUNEL staining in the bronchiolar epithelial cells (B). There was increased airway edema and congestion at 5 h with more consolidation by 24 h. D–F: Ki67 immunohistochemistry demonstrates increased staining in the peripheral lung (F) by 2 h with gradual increased staining until 24 h. Ki67 staining is increased in distal airways by 24 h (% airway epithelial cells positive) (G). Five-hour animals demonstrate increased Ki67 staining in peripheral lung tissue with minimal staining in the airways (I). Prominent staining in the terminal bronchioles by 24 h (J, inset) compared with controls (H, inset). Line = 50 μm. *P < 0.05 vs. control.

Fig. 3.

Pro-SPB and TTF-1 colocalization in Ki67 positive cells. A: club cell secretory protein (CCSP) mRNA decreases in the peripheral lung by 24 h after injury. B: prosurfactant protein B (red) is localized to vacuoles in the epithelial cells and colocalizes with Ki67 (green). C: TTF-1 protein (red) is present in most epithelial cells lining the terminal bronchioles and colocalizes with Ki67 positive cells by 24 h. *P < 0.05 vs. controls.

Fig. 4.

Proliferating cells of the terminal bronchioles are not basal cells. A: Western blot analysis of p63 in the peripheral lung tissue shows an increase in p63/β-actin by 30 min after mechanical ventilation that persisted for 24 h. Analysis includes 4 to 6 animals per group. B–G: immunohistochemistry of Ki67 (B and E, green) and p63 (C and F, red) demonstrate increased proliferation by 24 h (E) compared with controls (B). p63 signal is stable in the basal cells of the larger airways and mucosal glands and is not increased in the distal lung parenchyma. Colocalization of Ki67 and p63 (D and G), with DAPI nuclear staining, demonstrates the p63 positive basal cells are not proliferating. H–J: immunohistochemistry of Ki67 (green) and KRT5 (red) in 24 h animals confirms the basal cells are not proliferating. RLL, right lower lobe; RUL, right upper lobe.

Basal cell activation.

Airway basal cells expressed the proteins KRT5 and p63 at baseline (Fig. 4). KRT5 and p63 mRNA were detected in the trachea, left mainstem bronchi, and peripheral lung. p63 mRNA levels did not change in the peripheral lung tissue, but the RT-PCR Cq values were much higher (mean Cq 32.5) than in the left mainstem bronchi (Mean Cq 27.7), signifying there were not large numbers of basal cells in the peripheral lung tissue. Relative to p63, KRT5 mRNA did not change in the three airway regions across time (data not shown). Since p63 was consistently present in all cells of basal cell linage and p63 mRNA did not change in any region of the airways, we divided values of other basal cells markers by p63 quantification to account for sampling differences between animals and the number of basal cells within the segment (Fig. 5). Markers of basal cell activation KRT14 and KRT8 mRNA did not change in the trachea (Fig. 5, A and B) with mechanical ventilation, but increased in both the left mainstem bronchus of the lung (Fig. 5, C and D) and in the peripheral lung tissue (Fig. 5, E and F).

Fig. 5.

KRT14 and KRT8 mRNA responses in airways. mRNA expression of KRT14, KRT8, and p63 were assessed by RT-PCR. There were no differences seen in p63 between groups. To account for sampling differences in number of basal cells collected, a ratio of KRT14/p63 (A, C, and E) and KRT8/p63 (B, D, and F) was evaluated. A and B: KRT14 and KRT8 do not change over time in the trachea. C and D: KRT14 and KRT8 mRNA increase in the left mainstem bronchus. E and F: KRT14 and KRT8 increase in the peripheral lung tissue from the right lower lobe. *P < 0.05 vs. controls.

Notch activation and mucin production.

Notch pathway activation is involved in basal cell activation and transdifferentiation into goblet cells in mice (42). Similar to KRT14 and KRT8 production, a differential basal cell response was identified between the left main stem bronchus and the trachea. In the left mainstem bronchus, mRNA markers of the Notch pathway Hes 1, Hey 1, and Hey 2 did not change significantly after mechanical ventilation (Table 2). SPDEF, a goblet cell marker, also did not change. MUC5B increased slightly by 10 h after injury, but by 24 h after injury the level decreased to near baseline (Table 2). In contrast, in the trachea, the mRNA markers of Notch pathway Hey 1 and Hey 2 were elevated by 30 min after mechanical ventilation (Table 2). SPDEF, the transcription factor that identifies goblet cells, was elevated by 30 min and decreased to baseline levels by 24 h. MUC5B, the predominant mucin gene in sheep airways, increased by 30 min after injury and peaked by 10 h. Hes 1 did not to change after injury. There were no changes in Hes1, Hey1, or Hey2 mRNA in the peripheral lung tissue (data not shown).

Table 2.

Notch pathway and mucin production mRNA in larger airways

	Control	30 min	2 h	5 h	10 h	24 h
Left mainstem bronchus
Hes 1	1.0 ± 0.4	1.3 ± 0.2	1.0 ± 0.1	1.0 ± 0.2	1.0 ± 0.2	0.8 ± 0.1
Hey 1	1.0 ± 0.4	0.8 ± 0.01	1.4 ± 0.3	1.7 ± 0.3	1.6 ± 0.3	1.3 ± 0.1
Hey 2	1.0 ± 0.2	0.5 ± 0.1	0.5 ± 0.2	0.9 ± 0.4	0.5 ± 0.1	0.5 ± 0.1
SPDEF	1.0 ± 0.2	0.7 ± 0.2	1.4 ± 0.2	0.6 ± 0.1	1.1 ± 0.2	0.5 ± 0.02
Muc5B	1.0 ± 0.4	1.3 ± 0.2	1.7 ± 0.3	1.1 ± 0.3	3.8 ± 1.4*	1.6 ± 0.6
Trachea
Hes 1	1.0 ± 0.2	0.9 ± 0.1	1.3 ± 0.2	0.9 ± 0.1	0.9 ± 0.1	0.8 ± 0.1
Hey 1	1.0 ± 0.2	3.9 ± 1.3*	5.5 ± 2.7	2.0 ± 0.3*	2.5 ± 0.4*	4.7 ± 1.8*
Hey 2	1.0 ± 0.3	4.1 ± 1.4*	8.8 ± 4.8	1.1 ± 0.4	2.0 ± 0.7	2.7 ± 1.0
SPDEF	1.0 ± 0.3	2.2 ± 0.3*	2.3 ± 0.5*	1.8 ± 0.4	1.5 ± 0.3	1.1 ± 0.3
Muc5B	1.0 ± 0.3	2.9 ± 0.6*	3.8 ± 1.1*	3.3 ± 0.7*	6.7 ± 1.5*	3.1 ± 0.7*

Values are given as means ± SE.

^*P < 0.05 vs. control.

DISCUSSION

This study demonstrates that even a short period of large tidal volume ventilation (less than 10 min greater than 10 ml/kg) can cause significant injury to the preterm airways and initiates multiple repair mechanisms. Mechanical ventilation, in the absence of additional oxidative stress, caused lung inflammation and differential epithelial activation along the airways of preterm fetal sheep (Fig. 6). In the trachea, Notch pathway activation was associated with an increase in goblet cell mRNA. In the left mainstem bronchi, p63+ basal cells become KRT14 and KRT8 positive, likely due to epithelial growth factor receptor (EGFR) activation. Whereas the peripheral lung responded with proliferation of club (pro-SPB+, TTF-1+) cells. The airway cell proliferation (Ki67 +cells) response to mechanical ventilation, along with previously described smooth muscle cell hypertrophy, could contribute to the small airway injury prominent in children with a history of BPD (14, 16, 48). Elegant studies using transgenic mice and various injurious mechanisms (nathalene and bleomycin) have outlined cell-specific airway epithelial repair mechanisms (1, 39). This constellation of repair mechanisms cannot be fully explored in the human airways with their different distribution of cell types. Using a fetal sheep model system, with a similar lung distribution of epithelial cell types to the human lung (34), we demonstrate that some of the molecular pathways activated in adult murine injury models are activated in the preterm lung by mechanical ventilation (13). Understanding the repair pathways in the airways of preterm infants will be important if we want to decrease the reactive airway disease which is present in older survivors of BPD.

Fig. 6.

Summary of results based on anatomic location.

Mechanical ventilation at birth stretches the preterm airways and causes epithelial disruption and the release of inflammatory mediators (cytokines and heat shock proteins) (18, 21). More prolonged in utero ventilation (2 h) of fetal sheep previously demonstrated significant injury to bronchioles with increased collagen and elastin and decreased secondary crest formation (4, 5). These fetal sheep can recover from this lung injury and develop normal lung structures described at 15 days after injurious mechanical ventilation (4, 5). Since club cells are rare except in the most distal portions of the sheep airways, we hypothesized that the basal cells likely act as the progenitor cells for repairing airway injury throughout the bronchioles (13, 42, 44). In mouse trachea, epithelial injury requires basal cells (p63+, KRT5+) that become activated (expressing KRT14+, KRT8+) and transdifferentiate into multiple cell types to repopulate the epithelium (43, 44). In mice, up to 20% of basal cells are KRT14 positive in steady state (8). Our findings of increased KRT14 and KRT8 mRNA in the left mainstem bronchi tissue support the activation of basal cells in the preterm lung in response to injury (Fig. 6). The lack of epithelial disruption on hematoxylin and eosin or apoptosis, as demonstrated by no TUNEL staining, may indicate that the epithelium of the left mainstem was not severely injured, thus there were no epithelial-derived signals for proliferation of the basal cells (39). The injury from mechanical ventilation in these sheep was less severe and less uniform than seen in mouse models caused by toxin exposure (22, 29, 43). In the fetal lung, basal cell proliferation may occur with more severe epithelial injury or in the presence of oxidative stress.

The more distal portions of the mouse lung are repaired through transdifferentiation of multiple cell types (both club and ciliated cells) into a diverse epithelial layer (22). In certain situations, more differentiated cells can dedifferentiate into p63+, KRT5+ basal cells before progressing to lung repair (22). KRT5+, p63+ cells in the distal airspaces of mice exposed to influenza are essential for lung regeneration after influenza (29, 51). Lineage-negative epithelial progenitor cells in the distal regions of mouse lungs also express p63 and Krt5 and help with the lung repair (47). In the sheep model, the overall p63 protein level increased in the distal lung. Since p63 cells were not proliferating (minimal Ki67 colocalization) and there was no overall increase in p63 mRNA, the rapid increase in protein level within 30 min likely corresponded with the recruitment of inflammatory cells to the airways. The p53 transcription factors, of which p63 is a member, can modulate inflammation and be located in some inflammatory cell types (31, 33). Although the basal cells in the larger airways stained positive for p63 and Krt5, positive cells in the more distal airspaces were not very prominent. Markers of proliferation were in cells that colocalize with precursor for surfactant protein B and TTF-1 in the terminal bronchioles. The decrease in mRNA for CCSP suggests these club cells are differentiating and proliferating in response to the mechanical stretch. Unfortunately, we were unable to find an antibody for sheep CCSP and thus cannot definitely demonstrate that these cells were club cells. As with all mRNA data, without proper antibodies translation of mRNA cannot be confirmed. The lack of a TUNEL signal within the airway epithelium suggests the epithelium was not injured enough by the mechanical ventilation to trigger apoptosis. The increase in apoptosis, as measured by TUNEL assay, in the more peripheral lung could be due to cellular turnover caused by the maturational signals created by the mechanical ventilation (21). The normal lung development of the more peripheral lung tissue would involve apoptosis of some of the mesenchyme (49). The combination of club cell proliferation in the smallest airways and minimal apoptosis could contribute to the small airway disease found in extremely premature infants which survive BPD (14).

The trachea and the left mainstem bronchi have differential mRNA signals in response to mechanical stretch (Fig. 6). The responses in the bronchi are consistent with basal cell activation (42) and are likely due to the EGFR pathway activation. The mRNA levels for Notch pathways were not changed, nor were there consistent upregulation of mucin genes (SPDEF or MUC5B). Although increased AREG also contributes to the mucus cell metaplasia caused by naphthalene-induced lung injury (37), we did not find these changes in the left mainstem bronchi or more distal airways. The trachea responded with increased Notch ligand (Hey1, Hey2) mRNA and increased SPDEF and Muc5B mRNA. Notch signaling promotes a secretory phenotype and is essential for basal cells self-renewal in mice (11, 42). Notch pathway activation may result from multiple pro-inflammatory cytokines (IL-13, IL-6, IL-17) that are increased in sheep with mechanical ventilation (10, 25, 45). The repair pathway in the trachea is dependent on Notch signaling, with Notch2 driving the goblet cell hyperplasia (6, 10, 41). SPDEF is a transcription factor that regulates goblet cell hyperplasia in the lung and the intestine (15, 52). Stat6-dependant (allergic type) and Notch-dependent pathways increase SPDEF. SPDEF downregulation of the transcription factors FoxA2 and TTF-1 (nkx2.1) allows club cells to transdifferentiate into goblet cells, with minimal proliferation (Ki67) or apoptosis (cleaved caspase 3) (50). O'Reilly et al. (54) also ventilated fetal sheep and identified decreased apoptosis, decreased proliferation, and an increase in mucin containing cells in the airways with epithelial injury. There are likely subpopulations of basal cells that may be predisposed by Notch signaling to differentiate into secretory or ciliary cells (39).

Previous observations of progressive lung inflammation and maturation from a short period of mechanical ventilation were confirmed in this additional animal set (21). The decrease in the V40 over the progression of time and appearance of congestion within the lung tissue was consistent with lung inflammation and injury (18, 21). In response to the increased pro-inflammatory cytokines, the initial cells recruited to the lungs were neutrophils with a gradual appearance of monocytic cells and mature macrophages with vesicles. Although the fetal lamb has very few mature macrophages at birth, mechanical ventilation can cause the maturation of monocytes into Pu1+ alveolar macrophages in response to increased GM-CSF (21). It is unclear if these macrophages are involved in the inflammation (subtype M1) or part of the nonclassical repair process (subtype M2) (36).

Similar to our previous reports of EGFR elevation with 2 h of mechanical ventilation, a brief period of ventilation also stimulated EGFR ligand (AREG, EREG, and HB-EGF) expression in preterm lung (30). The absence of supplemental oxygen exposure in the fetal model demonstrates that stretch injury alone can activate the EGFR pathway. The mRNA increases in EREG and AREG were previously localized to cells with the characteristics of type 2 pneumocytes (15). In other model systems, EGFR ligands can be excreted by pulmonary cells in response to mechanical stretch and an autocrine activation of the EGFR receptor (3, 44, 45). EGFR ligands cause basal cell proliferation in human epithelial cultures (9) and is required for basal cells proliferation in mice (3). In asthma models, EGFR signaling mediates smooth muscle changes and airway hyperreactivity (30, 46). The activation of the EGFR pathway may be crucial to the development of BPD in preterm infants as EGFR is central for lung development. EGFR-deficient mice and rats exposed to anti-EGFR antibodies have decreased airway branching, decreased alveolarization, type II cell immaturity, and respiratory distress at birth (35, 38). Conversely, administration of EGF to preterm fetal rabbits increased lung maturation and increased type II alveolar epithelial cells (6). EGFR activation enhanced epithelial repair in sheep trachea (2). On a negative side, EGFR also participated in pulmonary fibrosis by modulation of TGF-α (12, 50). In the larger airways, EGFR activation also caused mucous hyperplasia and mediated the airway changes of acute and chronic asthma (30, 37, 46). EGFR activation can be protective or injurious and thus deserves further study in the setting of preterm lung injury.

A brief period of mechanical ventilation, in the absence of additional oxidative stress, can trigger lung inflammation, EGFR ligand production, and cellular proliferation. The preterm fetal sheep lung can repair epithelial injury, and this repair is likely due to a combination of the activation of basal cells in the bronchioles and club cells in the terminal bronchiole. Both EGFR and Notch pathways contribute to this repair from mechanical ventilation, though different responses occur in the trachea and smaller airways. The similarity of sheep lung with human lung in cellular distribution and maturation suggests that preterm infants will likely respond in a similar fashion to airway injury from mechanical ventilation. The activation of the basal cells (KRT14+, KRT8+) without proliferation would suggest that the short period of mechanical ventilation did not injure the larger p63 lined airways significantly. The proliferation of cells in the terminal bronchioles, likely club cells, could contribute to the small airway disease found in infants with history of BPD. Further research into the repair pathways of preterm infants is needed so that we might be able to augment this repair without causing the small airway disease found in adolescent survivors of extremely preterm birth.

GRANTS

Support for this study was provided by National Institutes of Health Grants R01-HD-072842 (to A. Jobe) and K08-HL-097085 (to N. Hillman).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

N.D., E.R., M.W.K., Y.M., A.H.J., and N.H.H. performed experiments; N.D., E.R., M.W.K., and N.H.H. analyzed data; N.D., E.R., and N.H.H. interpreted results of experiments; N.D., E.R., and N.H.H. prepared figures; N.D., E.R., A.H.J., and N.H.H. drafted manuscript; N.D., E.R., M.W.K., Y.M., S.G.K., A.H.J., and N.H.H. edited and revised manuscript; N.D., E.R., M.W.K., Y.M., S.G.K., A.H.J., and N.H.H. approved final version of manuscript; M.W.K., S.G.K., A.H.J., and N.H.H. conception and design of research.