Neutrophils promote alveolar epithelial regeneration by enhancing type II pneumocyte proliferation in a model of acid-induced acute lung injury

Abstract

Alveolar epithelial regeneration is essential for resolution of the acute respiratory distress syndrome (ARDS). Although neutrophils have traditionally been considered mediators of epithelial damage, recent studies suggest they promote type II pneumocyte (AT2) proliferation, which is essential for regenerating alveolar epithelium. These studies did not, however, evaluate this relationship in an in vivo model of alveolar epithelial repair following injury. To determine whether neutrophils influence alveolar epithelial repair in vivo, we developed a unilateral acid injury model that creates a severe yet survivable injury with features similar to ARDS. Mice that received injections of the neutrophil-depleting Ly6G antibody had impaired AT2 proliferation 24 and 72 h after acid instillation, which was associated with decreased reepithelialization and increased alveolar protein concentration 72 h after injury. As neutrophil depletion itself may alter the cytokine response, we questioned the contribution of neutrophils to alveolar epithelial repair in neutropenic granulocyte-colony stimulating factor (G-CSF)−/− mice. We found that the loss of G-CSF recapitulated the neutrophil response of Ly6G-treated mice and was associated with defective alveolar epithelial repair, similar to neutrophil-depleted mice, and was reversed by administration of exogenous G-CSF. To approach the mechanisms, we employed an unbiased protein analysis of bronchoalveolar lavage fluid from neutrophil-depleted and neutrophil-replete mice 12 h after inducing lung injury. Pathway analysis identified significant differences in multiple signaling pathways that may explain the differences in epithelial repair. These data emphasize an important link between the innate immune response and tissue repair in which neutrophils promote alveolar epithelial regeneration.

the acute respiratory distress syndrome (ARDS) is an acute inflammatory lung injury that can be triggered by several insults including pneumonia, sepsis, trauma, and aspiration (33). These varied insults result in a common histological pattern of inflammatory infiltrates, denudation of the basement membrane, and accumulation of a protein-rich edema fluid in the alveoli (8). The pathophysiology underlying this syndrome involves changes to barrier cells (5, 7, 63) that facilitate entry of inflammatory cytokines and leukocytes into the alveolar space (31). These changes impair gas exchange (7) and lead to hypoxemia, a clinical hallmark of the syndrome (5a).

Alveolar epithelial regeneration is a pivotal event in the resolution of ARDS that enables resorption of pulmonary edema fluid and improves gas exchange by restoring the alveolar architecture (48, 63). The alveolar epithelium is composed of alveolar type I pneumocytes (AT1) that are flat and facilitate gas exchange and alveolar type II pneumocytes (AT2) that secrete surfactant. Reepithelialization of the alveolus is dependent on new AT1 cells being generated (13, 42). Following epithelial destruction, the AT2 cell proliferates and differentiates into AT1 cells, thereby acting as a progenitor cell and promoting alveolar epithelial repair (9, 25). Other populations of AT1 progenitor cells have been identified, although our understanding of their role in the resolution of ARDS is incomplete (38, 60, 66).

Current conceptual models of ARDS suggest that neutrophils act as key mediators of alveolar epithelial damage by releasing oxidants, lipid mediators, and proteases (31, 62). While neutrophils are typically described as injurious in many organs and systems, several wound-healing models in organs other than the lung have demonstrated that neutrophils also promote epithelial repair (29, 41). One mouse model showed that neutrophil depletion significantly slowed the rate of skin wound closure (26). In this setting, oxidants, leukotrienes, and proteases are thought to promote sterility and beneficial tissue remodeling, which are essential for wound healing (41).

Recent studies have highlighted a potential role for neutrophils in supporting alveolar epithelial regeneration by promoting AT2 cell proliferation (1, 65). Specifically, in vitro studies have shown that neutrophil defensins promote alveolar epithelial cell proliferation (1) and in vivo studies have demonstrated that neutrophil transmigration from the vasculature into the alveolar space induces AT2 cell proliferation by activating β-catenin signaling (65). These studies portend a potentially important role for neutrophils in the restoration of the alveolar epithelium following ARDS by promoting AT2 cell proliferation. Furthermore, a recent study in mice showed that neutrophil accrual in the lung correlated positively with recovery from influenza, despite equivalent viral titers (40). Taken together, these studies suggest that traditional dogma may underestimate the importance of neutrophils in ARDS resolution and may explain, in part, why anti-inflammatory therapy fails to improve mortality in ARDS (16, 17, 46).

Although prior studies have demonstrated that neutrophils promote AT2 cell proliferation, the importance of neutrophils to alveolar epithelial regeneration following lung injury is largely unknown. To explore the role of neutrophils in alveolar epithelial repair we subjected neutropenic and control mice to unilateral acid-induced lung injury. This lung injury model produced a neutrophil-independent AT1 cell loss that was severe yet survivable (Figs. 1 and 3, A–C). Here we report that neutrophil depletion with a Ly6G antibody was associated with decreased reepithelialization of AT1 cells following acid-induced lung injury that was reflected in a prolonged alveolar protein leak among neutropenic mice. As previous studies have identified the AT2 cell as a key source of AT1 cells following lung injury (9, 42), we next sought to evaluate how the loss of neutrophils would impact AT2 cell proliferation. We found that neutrophil depletion was associated with significantly reduced AT2 cell proliferation, similar to observations made in a model of alveolar neutrophil chemotaxis (65). To account for possible confounding influences of neutrophil depletion, we sought to test our hypothesis in granulocyte-colony stimulating factor (G-CSF)−/− mice, which are known to be neutropenic, and found a similar positive correlation between neutrophils and alveolar epithelial repair. Furthermore, administration of recombinant G-CSF (rmG-CSF) to G-CSF−/− mice restored effective repair. While the mechanisms remain unknown, we hypothesized that neutrophils make multiple components that contribute to repair. Unbiased proteomic analysis of bronchoalveolar lavage (BAL) fluid from Ly6G antibody-treated and isotype antibody-treated mice revealed multiple pathways linking neutrophil-derived products to reparative cell signaling. This work is novel in that it explicitly studies the importance of neutrophils to alveolar epithelial repair in a model of injury and repair rather than implying their significance through chemotactic assays and in vitro models (1, 65).

Fig. 1.

Unilateral acid-induced lung injury. A: a PE-10 catheter (orange) is directed into the right lung via a 20-G angiocatheter (blue). Lung injury is induced with 2.5 μl/g 0.1 N hydrochloric acid. B: gross specimens confirm unilateral injury in the right lung.

Fig. 3.

Neutrophil depletion impairs epithelial recovery from acid-induced lung injury. C57BL/6 mice were treated with Ly6G or nonspecific IgG2a antibodies every 24 h beginning 1 day prior to undergoing acid-induced lung injury. A: lungs from nonspecific IgG2A and Ly6G antibody-treated mice obtained at 24 and 72 h following acid-induced lung injury. Lungs were fixed in 10% formalin under a fixed and constant pressure followed by paraffin-embedding and sectioning into 5-μm sections. H&E staining shows comparable lung injury at 24 h. The injury at 72 h after acid instillation appears to be more severe in the neutrophil-depleted mice, with a dense infiltrate and compromised alveolar architecture in the Ly6G-treated mice that was not observed in the mice that received the nonspecific IgG2A antibody. Scale bar = 50 μm. B: specimens from mice shown in A were stained with antibodies targeting the AT1-specific antigen podoplanin (red) and with DAPI to identify nuclei. Podoplanin staining of Ly6G- and isotype-treated mice has similar loss of epithelial cells at 24 h, but restoration of AT1 cells is not detected in the Ly6G antibody-treated group. Scale bar = 50 μm. Podoplanin staining of Ly6G- and isotype-treated mice have similar loss of epithelial cells at 24 h; however, restoration of AT1 cells is significantly reduced in the Ly6G antibody-treated group. Scale bar = 50 μm. C: specimens from mice shown in A and B were stained with antibodies targeting the AT1-specific aquaporin 5 (green) and demonstrated a pattern of loss and gain of AT1 cells; this is similar to what we observed with podoplanin staining in B. Scale bar = 100 μm. D: podoplanin staining of the uninjured left lung of a mouse 24 h after its right lung was injured with hydrochloric acid. E–G: BAL from mice treated with Ly6G and nonspecific IgG2A antibodies were analyzed at 12, 24, and 48 h after undergoing acid-induced lung injury. We observed no significant difference at 12 h; however, at 24 and 48 h after acid-induced lung injury the Ly6G antibody-treated mice did mice had a more prolonged alveolar protein leak compared with mice that received an isotype antibody. *P < 0.05 using Student's t-test. H: BAL samples from Fig. 3, E–G, were analyzed for IgM content using an ELISA kit. Comparison of Ly6G- and isotype antibody-treated mice showed a trend toward decreased IgM in the Ly6G-treated mice that was not statistically different using two-way ANOVA.

MATERIALS AND METHODS

Mice.

All mice were housed in specific pathogen-free conditions in an animal facility at the Children's Hospital of Philadelphia. All mouse protocols were approved by the Institutional Animal Care and Use Committee at the Children's Hospital of Philadelphia. GCSF−/− mice on a C57BL/6J background, strain 002398, from the Jackson Laboratory (Bar Harbor, ME) and WT C57BL/6J mice, strain 000664, from the Jackson Laboratory, aged 8–10 wk, were used for experiments. Both male and female mice from both genotypes were used in equal proportions.

Injury model.

Sedated mice were intubated with a 20-G angiocatheter from BD (Franklin Lakes, NJ) by a previously described technique (23). The mice were then placed in the right lateral recumbent position, and a polyethylene 10 (PE-10) catheter from Clay Adams (Parsippany, NJ) was directed into the right main stem bronchus while pressure was applied to the left lung. Injury was induced by instilling 2.5 μl/g of osmotically balanced 0.1 N HCl into the right lung through the PE-10 catheter. Immediately following acid instillation, all mice received a 250-μl bolus of ½ normal saline + 5% dextrose via an intraperitoneal injection. Mice then recovered for 4 h in a Plexiglas chamber, part no. AB-2 from Braintree Scientific (Braintree, MA). A 37° heating pad, part no. 39 DP from Braintree Scientific, was placed at the bottom of the chamber, and 100% oxygen was connected to the lower port of the chamber and pumped into the chamber at a rate of 10 l/min. After 4 h the mice were placed back into their cages exposed to room air, where they remained until they were euthanized.

Neutrophil depletion.

Neutrophil depletion was performed using a Ly6G antibody from Bio X Cell (West Lebanon, NH); a rat IgG2a control antibody from the same company was used as a control. Mice received a 270-μg injection of Ly6G antibody or IgG2a 24 h prior to acid instillation, on the day of acid instillation and every 24 h after that.

G-CSF repletion.

G-CSF repletion was performed using recombinant mouse G-CSF from BioLegend (San Diego, CA). Mice received 2-μg subcutaneous injections 48 h prior to injury and every 48 h after that for the duration of the experiment. Control mice received an equal volume of PBS.

Histological preparation.

Lungs were dissected and fixed under a constant pressure of 23 cmH₂O with 10% formaldehyde, dehydrated through a series of descending ethanol washes, embedded in paraffin wax, and sectioned at 5-μm sections (7 serial sections every 100 μm).

H&E staining.

Paraffin-embedded sections underwent deparaffinization, rehydration, and staining the Gemini AS from Thermo Fisher Scientific (Philadelphia, PA). Hematoxylin and eosin (H&E) were manufactured by Azer Scientific (Morgantown, PA).

Immunohistochemistry and confocal microscopy.

Following deparaffinization with xylene, the samples underwent antigen retrieval with a citric acid-based antigen unmasking solution from Vector Laboratories (Burlingame, CA) in a pressure cooker from Biocare Medical (Concord, CA). The following antibodies were used for immunohistochemistry: podoplanin clone 8.1.1 from the Developmental Studies Hybridoma Bank at the University of Iowa (Iowa City, IA) used at a concentration of 1:500, aquaporin 5 catalog no. AB92320 from Abcam used at a concentration of 1:100, prosurfactant protein C catalog no. AB3786 from EMD Millipore (Temecula, CA) used at a concentration of 1:200, and Ki67 catalog no. AB16667 from Abcam used at a concentration of 1:200. In cases where two primary antibodies were of rabbit origin (proSPC and Ki67), we used the TSA Biotin Staining Kit from Perkin-Elmer (Waltham, MA) according to the manufacturer's protocol to prevent cross-reactivity with the secondary antibodies. Confocal microscopy was performed using the Leica STED X3 microscope in the Cell and Developmental Biology Microscopy Core at the University of Pennsylvania. Images were subsequently processed using ImageJ.

ELISAs and protein assay.

BAL was obtained and cell counts were analyzed as previously described (52). Undiluted BAL and serum was tested using the mouse G-CSF ELISA kit from R&D Systems (Minneapolis, MN) and diluted BAL (1:200) was tested using the IgM ELISA kit from eBioscience/Thermo Fisher Scientific (Philadelphia, PA). Protein assays were done using the BCA Microplate BCA Protein Assay Kit from Thermo Scientific (Rockford, IL). All plates were read on the Spectra Max 250 spectrophotometer by Molecular Devices (Sunnyvale, CA).

Lung digestion and flow cytometry.

A single-cell suspension of lung tissue was generated by chopping lung tissue into fine pieces for 5 min followed by a subsequent enzymatic digestion in 480 U/ml collagenase type I from Life Technologies (Carlsbad, CA), 50 U/ml dispase from Collaborative Research (Bedford, MA), and 0.33 U/ml DNase I from Roche (Penzberg, Germany). Following an incubation in the digestive solution for 30 min the suspension was sequentially passed through 100 and 40 μM cell strainers from BD Falcon (Franklin Lakes, NJ). The red blood cells (RBC) were subsequently lysed with ammonium-chloride-potassium solution prepared in our laboratory using 8.3 g NH₄Cl, 1 g KHCO₃, and 200 μl of 0.5 M EDTA in 1 liter of sterile water. Following RBC lysis, the cells were resuspended in FACS buffer (2% FCS and 1 mM EDTA in PBS). Internal cells staining was accomplished by fixation and permeabilization with the Fox3P fixation and permeabilization kit from eBioscience/Thermo Fisher Scientific (Philadelphia, PA). The kit was used according the manufacturer's protocol with an 18-h incubation in the fixation/permeabilization buffer. The following antibodies were used for flow cytometry Ki67-APC from BioLegend (San Diego, CA), 7/4-PE from Serotec (Raleigh, NC), and pro-SPC from EMD Millipore (Temecula, CA) conjugated to a DyLight 488 fluorophore using a kit from Abcam (Cambridge, MA). Samples were read using the Accuri C6 machine from BD Biosciences (San Jose, CA). Data were analyzed using CFlow Plus software from BD Biosciences (San Jose, CA).

Sample preparation for mass spectroscopy.

One milliliter of each BAL was depleted of albumin and IgG using Proteoprep antibody depletion columns from Sigma-Aldrich (St. Louis, MO) following the manufacturer's instructions but sequentially adding 200 μl of BAL to the depletion column. After depletion, protein was precipitated with TCA. Control experiments indicated this resulted in ∼50% depletion of albumin (data not shown). After drying and solubilization in sample buffer, each sample was run into a 10% SDS-page gel, Coomassie stained, and excised in three equal segments. Each segment was further cut into 1-mm cubes, destained with 50% methanol/1.25% acetic acid, reduced with 5 mM dithiothreitol from Thermo Fisher Scientific (Waltham, MA), and alkylated with 20 mM iodoacetamide from Sigma-Aldrich (St. Louis, MO). Gel pieces were then washed with 20 mM ammonium bicarbonate from Sigma-Aldrich and dehydrated with acetonitrile from Thermo Fisher Scientific. Trypsin, 5 ng/ml in 20 mM ammonium bicarbonate, from Promega (Fitchburg, WI) was added to the gel pieces and proteolysis was allowed to proceed overnight at 37°C. Peptides were extracted with 0.3% triflouroacetic acid from Avantor (Center Valley, PA), followed by 50% acetonitrile. Extracts were combined and the volume was reduced by vacuum centrifugation.

Mass spectrometry analysis.

Tryptic digests were analyzed by LC-MS/MS on a QExactive HF mass spectrometer from Thermo Fisher Scientific coupled with an Ultimate 3000. Peptides were separated by reverse-phase (RP)-HPLC on a nanocapillary column, 75 μm ID × 25 cm 2 μm PepMap Acclaim column. Mobile phase A consisted of 0.1% formic acid from Thermo Fisher Scientific and mobile phase B of 0.1% formic acid/acetonitrile. Peptides were eluted into the mass spectrometer at 300 nl/min with each RP-LC run comprising a 90-min gradient from 10 to 25% B in 65 min, 25–40% B in 25 min. The mass spectrometer was set to repetitively scan m/z from 300 to 1,400 (R = 240,000) followed by data-dependent MS/MS scans on the 20 most abundant ions.

Protein data analysis.

Sequences were aligned with Sequest and analyzed in Scaffold. Differentially expressed proteins were analyzed using Ingenuity Pathway Analysis software from Qiagen (Hilden, Germany).

Statistics.

We used Microsoft Excel for Mac 2011 (Seattle, WA) and Prism 7 from GraphPad Software (La Jolla, CA). Means were compared by Student's t-test when comparisons were made between two groups at a single time point. When appropriate, we used two-way ANOVA. All t-tests were done Excel and two-way ANOVA testing was done in Prism 7. The distribution of data did not necessitate the use of nonparametric testing. P values <0.05 were considered significant. Statistical analysis of the proteomic data was performed using Scaffold. Peptides were analyzed both by sample and by quantity. To be considered significant, peptides from a protein were required to be present in at least two samples. Proteins determined to be more than two standard deviations removed from the line of identity were considered to be differentially expressed.

RESULTS

Unilateral acid instillation as a model of ARDS.

Clinically, ARDS is a severe lung injury that is not survivable without the support of mechanical ventilation (2, 56, 62). To mimic such a severe injury, we generated a model of unilateral lung injury. Using a unique method to selectively intubate the right lung (Fig. 1A) we were able to cause unilateral lung injury (Fig. 1B). We used 0.1 N hydrochloric acid (2.5 μl/g) to induce injury (defined as loss of AT1 cells from the lesion) that we hypothesized would not require neutrophils. Sparing one lung allowed the mice to survive what would otherwise be a lethal injury. Neutrophils (Fig. 2A) and protein-rich edema fluid (Fig. 3, E–G) accumulated in the alveoli following lung injury with peaks at 12 and 24 h, respectively. Furthermore, a pronounced loss of AT1 epithelial cells was detected within 24 h (Fig. 3, B and C). These data demonstrate that our lung injury model recapitulates several key elements of human ARDS.

Fig. 2.

Ly6G antibody decreases alveolar neutrophil accumulation in the lung following acid-induced lung injury. C57BL/6 mice were treated with Ly6G or nonspecific IgG2a antibodies every 24 h beginning 1 day before undergoing acid-induced lung injury. A: analysis of BAL taken from uninjured (0 h) and injured mice shows that neutrophil accumulation in the alveolar space peaked 12 h after acid-induced lung injury in mice that received the nonspecific IgG2A antibody. This response was significantly reduced by administration of a Ly6G antibody. **P < 0.01; ***P < 0.001 by two-way ANOVA. B: blood from mice treated with Ly6G antibody or a nonspecific IgG2A antibody was collected prior to inducing lung injury (0 h) and at multiple time points after inducing lung injury with 0.1 N HCl. Ly6G antibody significantly reduced circulating neutrophils in mice both before and following lung injury. *P < 0.05 by two-way ANOVA. C–E: lungs from mice that had been treated with a control IgG2A antibody or Ly6G antibody were digested into a single-cell suspension and analyzed by flow cytometry using the neutrophil-specific 7/4 antibody. C and D: forward scatter vs. side scatter plots from isotype antibody-treated (C) and Ly6G antibody-treated (D) mice. The blue arrow in C points to the area of the plot where we typically observe neutrophils. This population is absent in D. E: quantification of 7/4-positive cells shows a significant reduction of neutrophils in the Ly6G-treated mice relative to the isotype antibody-treated mice 24 h after acid-induced lung injury. ***P < 0.001 by Student's t-test.

Ly6G antibodies decrease the neutrophil response to acid-induced lung injury.

Intraperitoneal administration of Ly6G antibody not only depleted neutrophils from the circulation before injury (Fig. 2B) but also significantly decreased the neutrophil response to acid-induced lung injury compared with mice given an IgG2A isotype control antibody (Fig. 2, A–E). Twelve hours after injury, a time of peak neutrophil accumulation for mice given the isotype control antibody, there was an 89.1% (P = 0.0008) reduction in BAL neutrophils in the mice that received Ly6G antibody (Fig. 2A). The decrease in alveolar neutrophils persisted at 24 h when the Ly6G-treated mice had 70.5% (P = 0.038) fewer neutrophils than mice treated with the isotype antibody (Fig. 2A). These changes were reflected in peripheral blood counts at 12, 24, and 48 h after lung injury where the Ly6G had 84.2% (P = 0.0071), 74.0% (P = 0.052), and 68.9% (P = 0.014) fewer circulating neutrophils, respectively (Fig. 2B). While the decrease in circulating and BAL neutrophils argues strongly for significant depletion, we considered whether neutrophils might accumulate in the lung parenchyma without necessarily appearing in BAL. To address this possibility, we disaggregated the lungs of Ly6G and isotype control antibody-exposed mice and quantified the number of neutrophils in the right lung. Since the mice were depleted with Ly6G antibody, and the continued presence of the antibody might block binding of the Ly6G-detecting antibody, we used the antibody 7/4 (raised against murine neutrophils; does not detect Ly6G) as a detection antibody. As seen in Fig. 2, C–E, many fewer 7/4-staining neutrophils were detected in the Ly6g-exposed lungs than in isotype-exposed lungs. Taken together, these data support the contention that depletion with Ly6g antibody, while not absolute, nonetheless markedly diminishes neutrophils in the circulation, BAL, and lung tissue.

Neutrophil depletion impairs alveolar epithelial repair.

H&E staining of injured areas of lung 24 h after acid instillation shows accumulation of protein-rich edema fluid and septal thickening in the Ly6G and isotype antibody-treated mice (Fig. 3A). Immunofluorescent staining of lung tissue with the AT1-specific antibody directed against podoplanin also showed comparable AT1 cell loss in both neutropenic and control mice (Fig. 3B). Podoplanin was selected because it has previously been validated in the study of epithelial regeneration, including regeneration following acid-induced lung injury (19, 59). Nevertheless, we also confirmed our observation using a second method with an antibody targeting the AT1 cell-specific aquaporin 5 (Fig. 3C). H&E-stained lung sections 72 h after acid instillation demonstrate an inflammatory cellular infiltrate with organized alveolar structures in the isotype antibody-treated group. In contrast, the Ly6G-treated mice had an inflammatory infiltrate associated with the loss of distinct alveolar structures (Fig. 3A). Similarly, podoplanin staining showed partial reepithelialization with AT1 cells in the control mice, which was not observed in the neutropenic mice (Fig. 3B). A similar pattern was observed using aquaporin 5 antibody staining (Fig. 3C). Mirroring these differences in epithelial regeneration, we observed that BAL protein levels remained significantly elevated in neutropenic mice 48 h after inducing lung injury, 1.617 μg/μl, compared with control mice at the same time, 0.819 μg/μl (P = 0.019) (Fig. 3E). Interestingly, a similar trend was observed when we measured IgM (representing a large analyte, ∼1,000 kDa) levels in the BAL, although the data was not statistically significant (Fig. 3H).

AT2 cell proliferation is decreased in neutrophil-depleted mice.

Since neutrophil transmigration has been shown to promote AT2 cell proliferation in the uninjured lung (65), we examined whether neutropenia was associated with decreased AT2 cell proliferation in the response to injury, where the AT2 cell serves as a progenitor cell (9, 25). Twenty-four hours after inducing lung injury there was no significant difference in the number of AT2 cells per ×20 field between the neutropenic, mean = 60.67 cells per ×20 field, and control mice, mean = 63.71 cells per ×20 field (P = 0.52), suggesting that neutrophils do not influence AT2 cell survival. A significant increase in the percentage of proliferating AT2 cells, however, as measured by Ki67 staining (Fig. 4), was apparent even at 24 h. We observed that 5.73% of AT2 cells were Ki67 positive in the control group, whereas only 2.06% of AT2 cells were Ki67 positive in the neutrophil-depleted mice (P = 0.018). This decreased proliferation in the Ly6G-treated mice persisted at 72 h where 4.93% of AT2 cells were Ki67 positive compared with 15.58% in the control group (P = 0.0004). We observed a Ki67-positive rate in AT2 cells of less than 1% in uninjured mice (Fig. 4D).

Fig. 4.

Neutrophil depletion was associated with decreased AT2 cell proliferation following acute lung injury. A: lungs from mice treated with Ly6G or nonspecific IgG2A antibodies every 24 h, starting 24 h prior to inducing lung injury, were fixed in 10% formalin under fixed pressure and subsequently embedded in paraffin. We stained 5-μm sections using antibodies against proSPC and Ki67 alongside DAPI. Here we show representative confocal images of our staining scheme from a mouse that received a nonspecific IgG2a antibody 72 h after lung injury. Scale bar = 50 μm. B: high-power view of slides shown in 3A. Scale bar = 5 μm. C: for each treatment condition we obtained four ×20 images of alveolated lung tissue and quantified the number of proSPC+ cells (AT2 cells) and the number of Ki67+ AT2 cells at 24 and 72 h after lung injury. We observed a significant increase in the number of AT2 cells in the nonspecific IgG2A antibody-treated mice at 72 h relative to the Ly6G-treated mice. We also observed significant decreases in the number of proliferating cells at in the nonspecific IgG2A antibody-treated mice 24 and 72 h after being subjected to acid-induced lung injury. *P < 0.05 and ***P < 0.001 by Student's t-test. D: representative image of a lung section from an uninjured C57BL/6 mouse with staining for proSPC and Ki67 revealed that less than 1% of AT2 cells are proliferating in an uninjured mouse. We counted four ×20 fields per mouse in four mice and observed 9 Ki67+ cells of the 902 cells that were counted. Scale bar = 50 μm.

G-CSF−/− mice also have defective alveolar epithelial repair and decreased AT2 proliferation.

Neutrophil depletion itself can cause compensatory changes in mice, which may confound our data (22, 26). Therefore, we sought to validate our findings in a second model of neutropenia. G-CSF−/− mice were selected because they have a decreased neutrophil response to injury (11, 15, 45). Since transgenic mice may have secondary phenotypes from compensatory mechanisms that accumulate during development (10), we used G-CSF−/− mice that received rmG-CSF as controls.

Subcutaneous administration of rmG-CSF to G-CSF−/− mice started 48 h prior to acid instillation. BAL samples were obtained either prior to injury or 12 and 24 h after acid-induced lung injury. These time points for BAL analysis were chosen based on our observations that peak neutrophil accumulation occurs within the first 24 h following acid instillation (Fig. 2A). We observed that rmG-CSF accumulates in the lungs of G-CSF−/− mice only after initiation of lung injury (Fig. 5A) and that there is a significant 10.57-fold increase in neutrophil accumulation at 12 h (P = 0.0002) compared with G-CSF−/− mice that received PBS. Similar to neutrophil-depleted mice, G-CSF−/− mice had significantly impaired AT1 reepithelialization (Fig. 5E) 72 h after injury induction, a time when AT1 restoration is known to occur (Fig. 3, B and C). In mice repleted with G-CSF, alveolar epithelial type I cells were remarkably restored (Fig. 5E). Consistent with restoration of the alveolar permeability boundary, G-CSF−/− mice that underwent G-CSF repletion had an 80.4% (P = 0.001) decreased BAL protein concentration at 24 h (Fig. 5C) compared with G-CSF−/− mice that received PBS injections. A similar trend was observed when we measured IgM levels, although the data were not statistically significant (Fig. 5D).

Fig. 5.

G-CSF−/− mice also have impaired epithelial repair following acute lung injury. In this series of experiments we compared G-CSF−/− mice that received subcutaneous injections of G-CSF or PBS every 48 h starting 48 h prior to undergoing acid-induced lung injury. A–D: BAL was obtained from G-CSF replete and nonrepleted G-CSF−/− mice prior to injury (0 h) as well as 12 and 24 h after mice were exposed to acid-induced lung injury. A: we demonstrate that our injections resulted in successful repletion of G-CSF in the G-CSF−/− mice. B: we demonstrate that neutrophil accrual into the alveolar space is significantly reduced in G-CSF−/− mice that did not receive injections of rmG-CSF. C: we observe that BAL protein levels are higher in G-CSF−/− mice that did not receive rmG-CSF injections. For A–C, *P < 0.05, *P < 0.01, and ***P < 0.001 by two-way ANOVA. D: comparison of BAL from G-CSF-repleted mice to G-CSF mice that received PBS injections showed a trend toward decreased IgM in the rmG-CSF-treated mice that was not statistically significant when analyzed using two-way ANOVA. E: lung sections were fixed with 10% formalin under a constant pressure and subsequently paraffin embedded. Analysis of 5-μm sections using an anti-podoplanin antibody confirms more robust AT1 regeneration in G-CSF-repleted mice compared with G-CSF−/− mice that received PBS. Scale bar = 50 μm. F: for each treatment condition we obtained four ×20 images of alveolated lung tissue and quantified the number of proSPC+ cells (AT2 cells) and Ki67+ AT2 cells at 24 and 72 h after lung injury. We observed a significant increase in AT2 cells in rmG-CSF repleted-treated mice at 72 h. *P < 0.05 using Student's t-test. G: there were also significant differences in the number of proliferating AT2 cells in the rmG-CSF-repleted mice 24 and 72 h after being subjected to acid-induced lung injury. *P < 0.05 using Student's t-test. This phenomenon was confirmed using flow cytometry in Fig. 6.

Given our finding that neutrophil depletion was associated with decreased AT2 cell proliferation, we proceeded to perform a similar analysis in the G-CSF−/− model of neutropenia. Twenty-four hours after inducing lung injury there was no significant difference in the number of AT2 cells per ×20 field between the rmG-CSF-treated mice, mean = 75 cells per ×20 field, and PBS-treated mice, mean = 72.75 cells per ×20 field (P = 0.51), again suggesting that neutrophils do not influence initial AT2 cell survival (Fig. 5F). A significant increase in the percentage of proliferating AT2 cells, however, as measured by Ki67 staining (Fig. 5G), was apparent even at 24 h. We observed that 5.35% of AT2 cells were Ki67 positive in the rmG-CSF-treated group, whereas only 2.04% of AT2 cells were Ki67 positive in the PBS-treated mice (P = 0.035). This decreased proliferation in the PBS-treated mice persisted at 72 h where 2.48% of AT2 cells were Ki67 positive compared with 14.07% in the rmG-CSF-treated group (P = 0.044) (Fig. 5G). These findings were further verified by digesting G-CSF-replete and nonreplete G-CSF−/− lungs into a single-cell suspension that was stained with antibodies targeting pro-SPC and Ki67. We observed that a greater percentage of AT2 cells were Ki67 positive in G-CSF−/− mice treated with rmG-CSF compared with G-CSF−/− mice that received PBS (Fig. 6).

Fig. 6.

G-CSF repletion of G-CSF−/− mice enhances AT2 cell proliferation at 72 h following acid-induced lung injury. A and B: lungs from G-CSF−/− that received subcutaneous injections of rmG-CSF or PBS were collected 72 h after acid instillation. The blood was flushed out of the lungs using PBS prior to mechanically and enzymatically digesting the lung tissue. These single-cell suspensions were subsequently fixed and permeabilized and stained with antibodies targeting Ki67 conjugated to an APC fluorophore and proSPC conjugated to a DyLight 488 fluorophore. We used a flow cytometer to collect 1,000,000 events and analyzed the percentage of Ki67- and proSPC-copositive cells. A: here we show how the gates were set for each antibody using cells stained only with one antibody. B: we show that G-CSF-repleted mice have a marked increase in the percentage of Ki67-positive AT2 cells compared with cells from a G-CSF−/− mouse without repletion. Although it appears that there are more AT2 cells in the PBS-treated mouse that is likely because we collected a total of 1 million events from each mouse. The rmG-CSF mice have more inflammatory cells, which decreases the overall number of AT2 cells per million lung cells. For this reason, we report the percentage of AT2 cells that are Ki67 positive. Images shown for A and B are from a single experiment that was repeated two other times with similar results.

Proteomic analysis of BAL shows that neutrophils promote multiple regenerative pathways.

To interrogate the possible mechanisms through which neutrophils can promote alveolar epithelial regeneration, we performed an unbiased proteomic analysis comparing albumin-depleted BAL fluid taken from Ly6G antibody-depleted mice to that retrieved from mice given IgG2a isotype control antibody. We chose to analyze samples 12 h after injury since this is when we observed the greatest difference in BAL neutrophils between the two groups. Over 1,700 proteins in 1,500 groups were detected and mapped using Sequest and analyzed using Scaffold. A number of genes highly expressed in murine neutrophils (e.g., Ngp, Mmp9, S100a8, Retnlg, Sh3bgrl3, Lcn2, and Camp) were differentially detected in neutrophil-depleted samples, further supporting the effectiveness of depletion. More specifically, we found differences in proteins that have already been implicated in alveolar epithelial regeneration including Mmp9 (6, 21, 39), Mmp2 (39, 43), and Fgf1 (20, 55, 61), suggesting that neutrophils may modulate these pathways, in part, by influencing alveolar liquid components (Fig. 7). We have also included a list of proteins that were differentially upregulated (Table 1) or downregulated (Table 2) in mice given a nonspecific IgG2A antibody relative to mice that received the neutrophil-depleting Ly6G antibody. Future studies should focus on specific cell-signaling pathways as they may be important therapeutic targets.

Fig. 7.

Pathway analysis suggests that neutrophils activate multiple pathways related to cellular proliferation. Network analysis of data derived from an unbiased proteomic analysis of the BAL fluid 12 h after mice underwent acid-induced lung injury. Differentially expressed genes (2-fold or greater) were used to generate a series of networks that reflect, among others, inflammation, cell-cycle, and differentiation pathways. This pathway analysis suggests that neutrophils regulate key pathways such as Mmp9, Mmp2, and Fgf1, which have previously been described to enhance wound repair (3, 6, 30, 39, 43, 55, 61).

Table 1.

Proteins upregulated in neutrophil-replete compared to neutrophil-depleted BAL

Evc2*

vWF*

Zfp804b*

Mmp8*

Apoc2*

Mmp2*

Mmp9*

Etv5*

Col1a2*

Ctsd*

S100g*

Atp5d*

Rac2*

Mgp*

Ube2i*

Ngp

S100a8

Camp

Tmsb4x

Ptprf

Pi16

Odf4

Limch1

Sh3bgrl3

Igfbp1

Thbs4

Lap3

Calm1

Clic5

Mtpn

Mgam

Psmb2

Fgf1

Psma7

Sri

Fut11

Apon

Hdhd2

Masp1

Pla2g7

Cd9

Fis1

Coro1c

Fbxo38

Tceb1

Ppp2r4

Crk

Spp2

Fabp3

Psmb5

Cd93

Chia

Arpc5

Adh7

Serpina7

Retnlg

Proteins listed were detected in at least 2 of the 3 samples of the neutrophil-replete group. Proteins not detected in neutrophil-depleted BAL (e.g., only detected in the neutrophil-replete group) indicated with asterix (

^*). Samples expressed in both groups but upregulated in neutrophil-replete BAL selected based on >2 SD above the line of identity.

Table 2.

Proteins upregulated in neutrophil-depleted compared to neutrophil-replete BAL

Sdpr*
Calr*
Acad*
Fcho2*
Ganab*
Gpr45*
Gpt*
Adamts1*
Psma4*
Chmp4b*
Arpc3*
H2afy*
Fdps*
Fam81a*
Col18a1*
Mrc2*
Sorbs3*
Eno3*
Supt6h*
Mapk3*
Galk1*
Vcam1*
Oplah*
Cast*
Wdr95*
Cuta*
Cs*
Fkbp4*
Fmo3*
Pitpna*
Purb*
Acat1*
Atp5o*
Psmb7*
Nedd8*
Pmm2*
Iah1*
Umod*
Ptges3*
Tmod3*
Glul*
Pon3*
Cav1*
Atp6v1e*
Hk1*
Ide*
Psmd7*
Pea15*
Ppa1*
Atp5f1*
Uqcrc2*
Ppp1r2*
Sftpc*
Scrn2*
Myh10*
Cryab*
Rps20*
Slc39a4*
H2afx*
Nap1l1*
Ube2m*
Cpne3*
Cct5*
Ddx58*
Eef1b2*
Tars*
Ddah2*
H2afv*
Pura*
lS100a10*
Comt*
Rab1*
Dkk3*
Gmppb*
Txndc5*
Rap1a*
Banf1*
Nap1l4*
Tmem109*
Ppib*
Dpp9*
Ahsa1*
Ccdc129*
Ephx1*
Mstn*
Rab5c*
Ca8*
Mapk1*
Gsr*
Aox3*
Hspg2*
Irs4*
Sptbn1
Atp5a1
Ckm
Ptrf
Got2
Zyx
Alad
Acot2
App
Dld
Flnb
Tpm4
Cd14
Gss
Rps16
Mt2
Ang

Proteins listed were detected in at least 2 of the 3 samples of the neutrophil-depleted group. Proteins not detected in neutrophil-replete BAL (e.g., only detected in the neutrophil-depleted group) indicated with asterix (

^*). Samples expressed in both groups but upregulated in neutrophil-depleted BAL selected based on >2 SD above the line of identity.

DISCUSSION

Patient mortality in ARDS remains as high as 46% (2, 5a, 14, 32, 51, 54, 56) and the development of medications for this disease is a major unmet medical need (27). Based on the premise that neutrophils are the key mediators of alveolar damage [by releasing oxidants, lipid mediators, and proteases (31, 62)], therapeutic approaches have traditionally targeted limiting inflammation (16, 17, 46) while leaving the alveolar epithelium to recover on its own. Despite decades of clinical trials, this therapeutic approach has not demonstrated a convincing mortality benefit (16, 17, 46). More recent studies suggest that repair, not inflammation, may be the rate-limiting step to a clinical recovery (3, 34, 57). This notion underlies a new treatment paradigm that seeks to directly promote alveolar epithelial regeneration (47, 57).

Recent studies have established an important link between inflammation and repair, with notable overlap in cell signaling between both processes (35, 36, 50, 58). With regard to neutrophils, transmigration across the alveolar epithelium was recently shown to induce AT2 cell proliferation in vivo (65). This finding is supported by similar data evaluating neutrophil defensins in vitro (1). Moreover, clinical reports indicate patients with profound neutropenia develop ARDS (44, 62). These data suggest that the function of neutrophils in ARDS is not fully understood. Therefore, current conceptual models may underestimate the importance of neutrophils in ARDS resolution and may explain, in part, why anti-inflammatory therapy does not improve mortality in ARDS (16, 17, 46).

The experiments detailed in this report further explore the idea that inflammation promotes alveolar epithelial repair. To study the role of the neutrophil, in particular, we needed to use a lung injury model that was cytotoxic to the epithelium independent of neutrophils and would also allow mice to survive long enough to study repair. Therefore, we developed a novel method for instilling hydrochloric acid into the right lungs of mice, based on earlier reports (4, 64). As shown, this injury model results in a histological injury pattern similar to the diffuse alveolar damage pattern seen in ARDS, including septal thickening, cellular infiltration, accumulation of protein-rich edema fluid, and denudation of the basement membrane (5, 8).

Using a Ly6G antibody to induce neutropenia we observed that neutrophils do not impact acid-induced epithelial lung injury as assessed histologically or by measuring BAL protein concentration. Since formal lung injury grading systems use neutrophils in their scoring (12, 49), we have focused here specifically on loss and repair of the critical AT1 cell boundary. However, we did observe comparable BAL protein levels at 12 h, a time of peak inflammation in the control mice. Despite having similar initial injuries, we observed several differences at later time points in alveolar epithelial repair including decreased restoration of AT1 cells and prolonged alveolar protein leak in the neutrophil-depleted mice relative to the control mice. BAL IgM concentrations, although markedly elevated, were not statistically different between the two groups, suggesting that large pores through which IgM passes may be regulated differently than other permeability barriers at the time points that we studied.

These findings are echoed by other groups that have shown that neutropenia is associated with decreased epithelial repair (24, 26, 29, 53) outside the lung as well as other groups that have demonstrated that neutrophil products can promote AT2 cell proliferation (1, 65). Despite these observations, this is the first study to investigate the potential role of neutrophils in epithelial repair following a model of lung injury. Our quantification of AT2 cell proliferation shows that neutropenia is associated with decreased AT2 cell proliferation, which we believe underlies the differences in AT1 cells seen 72 h after acid-induced lung injury.

These findings are further supported by an unbiased proteomic analysis that identified neutrophil-derived products in the BAL of isotype-exposed lungs compared with Ly6g-exposed lungs. These products, as well as non-neutrophil-derived proteins detected in the same sample, represent candidate molecules to illuminate pathways related to AT2 proliferation and differentiation. In particular, we wish to highlight the potential roles of Mmp9 and Mmp2, and Fgf1, all detected in BAL from neutrophil-replete mice. Mmp9 and other Mmp2 have been suggested to mediate repair by altering basement membrane constituents (6, 30, 37, 39, 43). Fgf1 is a also a candidate to drive ATII proliferation (18, 20, 55, 61). We recognize that pathway analysis is limited and that it may highlight pathways and molecules that are not involved. Nonetheless, we believe that hypotheses regarding mechanisms of alveolar epithelial repair may be facilitated by this information.

In summary, data generated using a novel unilateral lung injury model show that neutropenia is strongly associated with decreased AT2 cell proliferation and AT1 regeneration in response to acute lung injury. Although other groups have suggested that such a relationship exists (65), we are the first to formally test this hypothesis in a model of injury and repair using two models to decrease the neutrophil response to injury. Furthermore, we have used an unbiased proteomic analysis to identify signaling pathways that are altered by neutrophil depletion. The components of these pathways, Mmp9, Mmp2, and Fgf1, among many others, may promote alveolar epithelial repair and merit further investigation. In the meantime, as the innate immune response appears to promote alveolar epithelial regeneration, these data provide further evidence for considering approaches other than anti-inflammatory medications in the treatment of patients with ARDS.

GRANTS

This research was funded with support from the following grants: 1R01AI099479, 5R01HL10583402, 5T32HL07586, and 1F32HL131079-01.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

A.J.P., Y.L., J.M., L.G., K.H., J.B., M.F.B., E.E.M., and G.S.W. conceived and designed research; A.J.P., Y.L., N.D., L.G., L.S., W.Z., H.M., A.S., K.B., S.H.S., and G.S.W. performed experiments; A.J.P., Y.L., J.M., L.G., W.Z., M.F.B., S.H.S., and G.S.W. analyzed data; A.J.P., Y.L., J.M., L.G., J.B., M.F.B., E.E.M., and G.S.W. interpreted results of experiments; A.J.P. and G.S.W. prepared figures; A.J.P. and G.S.W. drafted manuscript; A.J.P., K.H., and G.S.W. edited and revised manuscript; A.J.P. and G.S.W. approved final version of manuscript.