Research ArticleLUNG INJURY

Neutrophil transfer of miR-223 to lung epithelial cells dampens acute lung injury in mice

See allHide authors and affiliations

Science Translational Medicine  20 Sep 2017:
Vol. 9, Issue 408, eaah5360
DOI: 10.1126/scitranslmed.aah5360

Starting an intercellular conversation

In a new study, Neudecker et al. show that transfer of microRNA-223 (miR-223) from neutrophils to lung alveoli helps to dampen lung inflammation and promotes the resolution of ventilator-induced lung injury in mice. The authors suggest that neutrophils secrete microRNAs in microvesicles that are then taken up by alveolar epithelial cells. They show that miR-223–deficient mice are prone to lung injury, whereas overexpression of miR-223 is protective.

Abstract

Intercellular transfer of microRNAs can mediate communication between critical effector cells. We hypothesized that transfer of neutrophil-derived microRNAs to pulmonary epithelial cells could alter mucosal gene expression during acute lung injury. Pulmonary-epithelial microRNA profiling during coculture of alveolar epithelial cells with polymorphonuclear neutrophils (PMNs) revealed a selective increase in lung epithelial cell expression of microRNA-223 (miR-223). Analysis of PMN-derived supernatants showed activation-dependent release of miR-223 and subsequent transfer to alveolar epithelial cells during coculture in vitro or after ventilator-induced acute lung injury in mice. Genetic studies indicated that miR-223 deficiency was associated with severe lung inflammation, whereas pulmonary overexpression of miR-223 in mice resulted in protection during acute lung injury induced by mechanical ventilation or by infection with Staphylococcus aureus. Studies of putative miR-223 gene targets implicated repression of poly(adenosine diphosphate–ribose) polymerase–1 (PARP-1) in the miR-223–dependent attenuation of lung inflammation. Together, these findings suggest that intercellular transfer of miR-223 from neutrophils to pulmonary epithelial cells may dampen acute lung injury through repression of PARP-1.

INTRODUCTION

Acute lung injury manifests as acute respiratory distress syndrome (ARDS) in patients (1). Its clinical features include acute hypoxemic respiratory failure and bilateral pulmonary infiltrates that are not attributable to left heart failure (13). During acute lung injury, inflammatory cells, particularly polymorphonuclear neutrophils (PMNs), come into close contact with lung alveolar epithelial cells.

Many research studies have provided insights into intercellular communications regulating neutrophil activation and pulmonary transmigration during acute lung injury (4). These communications include paracrine cross-talk between neutrophils and lung parenchymal cells. For example, previous studies have shown that PMNs release extracellular nucleotides (for example, adenosine triphosphate) that are converted into adenosine, which dampens pulmonary epithelial inflammation (5, 6) and improves fluid transport during acute lung injury (7, 8). Here, we investigated whether PMNs could participate in intercellular communication with lung alveolar epithelial cells through microvesicle-dependent exchange of microRNAs (miRNAs) (9).

miRNAs constitute a family of short noncoding RNA molecules of 20 to 25 nucleotides in length that regulate gene expression at the posttranscriptional level (10). Bioinformatic predictions indicate that more than 60% of all mammalian genes are potentially regulated by miRNAs (11). Although the investigation of functional miRNA target genes has identified putative regulatory functions for miRNAs (12), little is known about the repression of inflammatory genes by miRNAs during acute lung injury. Here, we investigated whether PMN–epithelial cell cross-talk during acute lung inflammation could include the exchange of miRNAs (12).

RESULTS

miR-223 can be transferred from neutrophils to pulmonary epithelial cells

Previous studies have indicated that neutrophil (PMN)–epithelial cell cross-talk can dampen inflammation (13). On the basis of these findings, we hypothesized that during neutrophil–epithelial cell interactions, genetic information in the form of miRNAs could be transferred from PMNs to pulmonary epithelia. To test this hypothesis, we set up an in vitro coculture system of human primary alveolar epithelial cells (HPAEpiC) with freshly isolated human PMNs, where both cell types were separated by a membrane with a pore size of 0.4 μm, preventing direct cell-cell contact (Fig. 1A). After 6 hours of coincubation, we washed the alveolar epithelial cells, isolated miRNAs, and performed a targeted expression analysis of miRNAs known to be expressed in human PMNs (14). We observed a robust (more than 100-fold) selective increase in human miR-223 (hsa–miR-223) (Fig. 1B). Confirmatory studies demonstrated that this increase occurred in a time-dependent manner in human primary alveolar epithelial cells (HPAEpiC) and in the pulmonary epithelial cell line Calu-3 after coculture with human PMNs (Fig. 1C and fig. S1A). Analysis of baseline amounts of hsa–miR-223 in pulmonary epithelia displayed very low expression of miR-223 [cycle threshold (Ct) value, 29.14 ± 0.61; culture medium Ct value, 34.52 ± 1.12], whereas PMNs demonstrated high expression of hsa–miR-223 (Ct value, 7.95 ± 0.40) (fig. S1B). Also, the expression of hsa–miR-223 in HPAEpiC was not inducible by various stimuli tested including exposure to N-formyl-Met-Leu-Phe (fMLP) or lipopolysaccharide (LPS) (Fig. 1, D and E). In contrast, hsa–miR-223 was found to be about 20-fold lower after coculture of PMNs with human microvascular endothelial cell–1 (HMEC-1) (15, 16) than coculture with human pulmonary epithelial cells (Calu-3) (fig. S1C). To test whether the hsa–miR-223 detected in human pulmonary epithelial cells after coculture was functional, we performed coculture studies with human pulmonary epithelial cells (Calu-3) that were previously transfected with a luciferase reporter carrying a miR-223 target sequence. Significant decreases (P < 0.05) in luciferase activity in Calu-3 after coculture indicated that hsa–miR-223 was functional after coculture (Fig. 1F). To provide additional evidence that increases in pulmonary epithelial cell miR-223 after coculture were due to PMNs, we used a murine coculture system that allowed us to study miR-223–deficient PMNs (Fig. 1G). In this system, a murine pulmonary epithelial cell line (MLE-12) was cocultured with murine PMNs derived from either wild-type or miR-223−/y mice. The miR-223 gene is located on the X chromosome; therefore, the knockout mice were hemizygous for miR-223 (miR-223−/y). The deficiency in miR-223 was confirmed by analyzing miR-223 in murine neutrophils from miR-223−/y mice compared to wild-type mouse neutrophils (fig. S1, D and E). Analyses of murine miR-223 (mmu–miR-223) after 6 hours of coculture with wild-type PMNs revealed a significant increase (P < 0.05), whereas no alteration in epithelial cell mmu–miR-223 expression was observed after coculture with murine PMNs derived from miR-223−/y mice (Fig. 1H). Moreover, a comparison of miR-223 shuttling in the coculture system comprising murine alveolar epithelial cell type I or II (AT-II–like cells, MLE-12 cell line; AT-I–like cells, E-10 cell line; Fig. 1I) indicated that miR-223 transfer predominantly occurred from neutrophils to AT-II cells. Together, these findings indicate that miR-223 can be transferred from PMNs to pulmonary epithelial cells under coculture conditions.

Fig. 1. Transfer of miR-223 during neutrophil-epithelial cell interactions.

(A) Coculture setup for human neutrophils (PMNs) and human pulmonary epithelial cells. (B) Expression of miRNA in human epithelial cells after coculture of HPAEpiC with activated human PMNs (means ± SEM; n = 4). (C) hsa–miR-223 expression after coculture of HPAEpiC with activated human PMNs (means ± SEM; n = 6). (D and E) Expression of miR-223 in HPAEpiC after exposure of HPAEpiC cells to (D) fMLP or (E) LPS [means ± SEM; n = 2 for control (Ctr) group or n = 3 for fMLP and LPS groups]. (F) Epithelial cell miR-223 target vector luciferase activity after coculture of activated human PMNs with transfected pulmonary epithelial cells (Calu-3); data are normalized to control vector activity and compared to no coculture (means ± SEM; n = 3 independent experiments). (G) Setup for murine coculture. (H) mmu–miR-223 expression in mouse pulmonary epithelial (MLE-12) cells after coculture with activated murine PMNs derived from wild-type (WT) or miR-223−/y mice (means ± SEM; n = 11 for the control group, n = 6 for mouse WT PMNs, and n = 7 for miR-223−/y PMNs). (I) mmu–miR-223 expression in MLE-12 cells or E-10 cells after coculture with activated murine PMNs (means ± SEM; n = 7 for MLE-12 control group or n = 8 for all other conditions) [*P < 0.05, Student’s t test or analysis of variance (ANOVA)]. n.s., not significant.

miR-223 release from PMNs is activation-dependent

Next, we examined the supernatant from PMNs for miR-223 content. PMNs released high amounts of miR-223 into the supernatant (Fig. 2A). Isolation of microvesicles from the supernatant from PMNs via ultracentrifugation showed a high concentration of miR-223 (Fig. 2B). In addition, incubation of PMN supernatants with HPAEpiC was associated with increased miR-223 in HPAEpiC, whereas this response was almost completely abolished when using PMN supernatants that lacked the microvesicle fraction (Fig. 2C). Besides the secretion of microvesicles, activation of PMNs induced the release of decondensed chromatin and granular contents into the extracellular space (17). These structures, called neutrophil extracellular traps (NETs), have antimicrobial properties but have also been implicated in inflammation. We therefore asked whether the release of NETs (NETosis) was involved in the secretion of miR-223 by PMNs. As shown previously, we observed that activation of PMNs was associated with increased NETosis (fig. S2A). To address the possibility that increased NETosis could also be associated with increased release of miR-223, we repeated these studies in the presence of a NETosis inhibitor. Effective inhibition of NETosis did not affect miR-223 release, thereby providing evidence that NETosis was not a major pathway mediating extracellular miR-223 release (fig. S2, A and B). To provide further evidence for PMN–epithelial cell miRNA transfer, we transfected a human neutrophil–like cell line (HL-60) with cel–miR-39, a miRNA only present in Caenorhabditis elegans, and assessed cel–miR-39 expression in human pulmonary epithelial cells (Calu-3) after coculture. We found a robust increase in cel–miR-39 in human pulmonary epithelial cells (Calu-3) (fig. S2C). Moreover, the addition of monodansylcadaverine (a known inhibitor of clathrin-mediated endocytosis) to the supernatant partially attenuated the alveolar epithelial cell uptake of miR-223–containing microvesicles, thus implicating clathrin-mediated endocytosis in miR-223 transfer (Fig. 2D). Consistent with our work showing PMN transfer of miR-223 into AT-II cells (Fig. 1I), previous studies have demonstrated that clathrin-mediated transport is functionally important in AT-II cells (18, 19). Together, these findings suggest that miR-223–containing microvesicles may be released from PMNs after PMN activation and then could be transferred into alveolar epithelial cells.

Fig. 2. PMN-dependent transfer of miR-223 to human pulmonary epithelial cells.

(A) hsa–miR-223 release into human PMN–derived supernatants was compared to culture medium only (means ± SEM; n = 4 in the culture medium–only group and n = 6 in the PMN-derived supernatant group) after human PMN activation. (B) hsa–miR-223 in PMN supernatant–derived microvesicles compared to medium only (means ± SEM; n = 4). (C) Expression of miR-223 in HPAEpiC cells after incubation with PMN-derived or control supernatants containing microvesicles (+MV) or depleted of microvesicles (−MV) (means ± SEM; n = 7). (D) hsa–miR-223 expression after coculture of activated human PMNs with HPAEpiC cells in the presence or absence of the endocytosis inhibitor monodansylcadaverine (MDC) (means ± SEM; n = 8). (E) Pulmonary expression of mmu–miR-223 in a mouse model of VILI (means ± SEM; n = 6 mice per group). (F) mmu–miR-223 expression in isolated mouse AT-II cells after VILI (means ± SEM; n = 5 mice in the control group and n = 6 mice in the VILI group). (G) mmu–miR-223 expression in AT-II cells isolated from a mouse bone marrow (BM) chimera (WT→miR-223−/y and miR-223−/y→WT) after exposure to VILI (means ± SEM; n = 5 mice in the WT→miR-223−/y VILI group and n = 4 mice in all other groups). (H) mmu–miR-223 expression in murine lung tissue during VILI after antibody-mediated depletion of PMNs (means ± SEM; n = 5 mice in the control-treated group, n = 4 mice in the 1A8 antibody-treated group, and n = 7 mice in the control-treated VILI group). (I) Intravital microscopy of HL-60 cells, transfected with a dye-labeled miRNA mimic (red) and labeled with calcein dye (green), after intravenous injection into mice ventilated for 30 min (top) or after 90 min of high tidal ventilation to induce VILI (bottom) (two representative images; n = 3 mice). (J) Immunohistochemical (IHC) costaining of AT-II cells [fluorescein isothiocyanate (FITC)–pro–surfactant protein C (SPC); green] and Qtracker-labeled and intravenously injected HL-60 cells (Qtracker-655; red) in murine lung tissue after induction of VILI (one representative image; n = 3 mice). (K) hsa–miR-223 expression in BAL fluid derived from patients with ARDS or healthy controls (median and 25 to 75% percentiles; n = 8 controls and n = 55 ARDS patients; Table 1) (*P < 0.05, Student’s t test, ANOVA, or nonparametric analysis Mann-Whitney test).

PMNs are required for increased pulmonary miR-223 during acute lung inflammation

Given that neutrophils are in close contact with pulmonary epithelial cells during lung inflammation, we next examined mmu–miR-223 expression during acute lung inflammation in mice. For this purpose, we used a mouse model where lung injury was induced by mechanical ventilation [ventilator-induced lung injury (VILI)] (7, 2022). We found that mmu–miR-223 in whole lungs was significantly increased after induction of acute lung inflammation in mice (P < 0.05; Fig. 2E). This increase was associated with an increase in PMN-derived (CD11b+Ly6G+) microvesicles (fig. S3, A and B). In line with a possible transfer of miR-223 from PMNs to alveolar epithelial cells during acute lung inflammation, we found that mmu–miR-223 in isolated mouse alveolar epithelial cells was elevated after induction of acute lung inflammation (Fig. 2F), whereas miR-223 expression in isolated mouse pulmonary endothelial cells remained the same (fig. S3C). To generate mice chimeric for miR-223, wild-type mice underwent sublethal radiation and were transplanted with miR-223−/y mouse bone marrow and vice versa. Studies of these bone marrow chimeric mice revealed that increased mmu–miR-223 in mouse alveolar epithelial cells during pulmonary inflammation was predominantly dependent on the presence of mmu–miR-223 in bone marrow–derived cells (Fig. 2G and fig. S4, A to C). Similarly, increased mmu–miR-223 expression during acute lung inflammation was attenuated after antibody depletion of PMNs (Fig. 2H and fig. S4, D and E). Although the antibody treatment resulted in significant neutrophil depletion, it did not affect monocytes (fig. S4E). Consistent with these findings, miR-223 expression was more than 25-fold higher in PMNs than in alveolar macrophages (AMφ) (fig. S5A), and transfer of miR-223 from bone marrow–derived macrophages (BMDMφ) to mouse pulmonary epithelial cells was not detectable during coculture (fig. S5B). Moreover, induction of miR-223 expression during acute lung inflammation induced by mechanical ventilation remained intact after macrophage depletion with clodronate (fig. S5C). To study PMN-dependent miRNA release during acute lung injury in mice, we examined animals that had been injected with PMN-like cells (HL-60) that had been transfected with a synthetic fluorescently labeled miRNA. Intravital microscopy through a transthoracic window revealed evidence for release of miRNAs from HL-60 cells that were in close proximity to mouse alveolar epithelial cells during acute lung injury induced by mechanical ventilation (Fig. 2I). In addition, costaining for AT-II cells (Fig. 2J, green) in murine lung tissue sections from ventilated mice that had been injected with labeled PMN-like cells (HL-60; Fig. 2J, red) suggested that there may also have been transfer of cytoplasm from the PMN-like cells in vivo. Moreover, we also observed that patients with ARDS (Table 1) showed increased hsa–miR-223 in their bronchoalveolar lavage (BAL) fluid compared to healthy control patients (control group mean Ct value for miR-223, 26.61; mean Ct value for control ce–miR-39, 23.32; ARDS group mean Ct value for miR-223, 25.35; mean Ct value for control ce–miR-39, 24.23) (Fig. 2K). Together, these findings indicate that PMN-dependent production of mmu–miR-223 contributed to increased miR-223 expression in injured mouse lungs after mechanical ventilation and suggest that miR-223 was transferred to mouse alveolar epithelial cells during acute lung inflammation.

Table 1. Characteristics of patients providing BAL fluid samples.

hsa–miR-223 analysis was assessed in BAL fluid collected from 55 patients with ARDS within the first 7 days after diagnosis. Demographic data and APACHE II (Acute Physiology and Chronic Health Evaluation) clinical scores are shown. BAL samples from eight healthy nonsmokers (five males and three females) served as controls. n/a, not applicable.

View this table:

PARP-1 is a miR-223 target gene in pulmonary epithelial cells

We next performed studies to identify miR-223 target genes in pulmonary epithelial cells. For this purpose, we used lentiviral transduction to generate a pulmonary epithelial cell line (Calu-3) that overexpressed miR-223 (Fig. 3, A and B). A whole-genome microarray comparing control-transfected Calu-3 cells with Calu-3 miR-223–overexpressing cells identified 83 genes with decreased expression in the miR-223–overexpressing cell line (table S1). Among genes containing a miR-223 seed match (TargetScan, miRanda, MicroCosm, and miRDB), the gene most robustly repressed after miR-223 overexpression was poly[adenosine diphosphate (ADP)–ribose] polymerase–1 (PARP-1). PARP-1 is known to facilitate inflammatory responses by activating transcription factors such as nuclear factor κB (NF-κB) and activator protein–1 (AP-1). PARP-1 drives inflammation-relevant gene expression via various proinflammatory cytokines, adhesion molecules, and enzymes involved in inflammation. Previous studies demonstrated a detrimental role for PARP-1 during acute lung injury induced by mechanical ventilation (23). In line with studies from the cancer field (24), we confirmed repression of PARP-1 transcripts (Fig. 3C) and protein (Fig. 3, D and E) after miR-223 overexpression. We next utilized two luciferase reporter plasmids with one containing the wild-type 3′ untranslated region (3′UTR) of PARP-1 and the other carrying a mutated version where specific binding of miR-223 to the 3′UTR was abrogated (Fig. 3F). Transfection of these plasmids into human pulmonary epithelial cells confirmed repression of the wild-type PARP-1 plasmid after miR-223 overexpression, which did not occur with the mutated plasmid (Fig. 3, G and H). A coculture system comprising human PMNs and human pulmonary epithelial cells (Calu-3) (Fig. 1A) revealed repression of PARP-1 transcripts in the pulmonary epithelial cells (Fig. 3I). Together, these findings identify PARP-1 as a miR-223 target gene.

Fig. 3. PARP-1 is a miR-223 gene target in human pulmonary epithelial cells.

(A) Baseline hsa–miR-223 expression in lentiviral-transduced, human Calu-3 (lenti-223) cells compared to control Calu-3 (lenti-ctr) cells (means ± SEM; n = 4). (B) Epithelial cell miR-223 target vector luciferase activity in transfected human Calu-3 cells compared to control vector activity (means ± SEM; n = 6). (C) Analysis of PARP-1 transcripts in lentiviral-transduced human Calu-3 cells compared to control Calu-3 cells using real-time reverse transcription polymerase chain reaction (RT-PCR) relative to the housekeeping gene β-actin (means ± SEM; n = 8). (D and E) Western blot analysis of PARP-1 protein in lentiviral-transduced human Calu-3 cells compared to control Calu-3 cells (one representative blot of three is shown; quantification by densitometry, n = 3). (F) Schematic of luciferase reporter plasmids showing the binding sequence between PARP-1 3′UTR and hsa–miR-223. (G and H) Luciferase activities of PARP-1 3′UTR WT plasmid or mutated (MUT) plasmid compared to control 3′UTR plasmid after transfection into lentiviral-transduced human Calu-3 cells (lenti-ctr or lenti-223) (means ± SEM; n = 12). (I) PARP-1 transcripts in the human Calu-3 cell line after coculture with human PMNs measured by real-time RT-PCR relative to the housekeeping gene β-actin (means ± SEM; n = 6) (*P < 0.05; Student’s t test).

miR-223−/y mice develop severe lung inflammation during acute lung injury

To address the functional role of miR-223 during pulmonary inflammation, we next performed genetic studies in miR-223−/y mice (25). We induced acute lung injury using mechanical ventilation in miR-223−/y mice or wild-type control mice matched for age, gender, and weight. Consistent with a functional role for miR-223 in repressing inflammatory gene expression, we found that, compared to control mice, miR-223−/y mice showed increased permeability of lung alveolar cells (Fig. 4A). The miR-223−/y mice also showed greater lung inflammation as assessed by measurement of bronchoalveolar myeloperoxidase (Fig. 4B), bronchoalveolar cytokines (Fig. 4, C and D), histological tissue injury (Fig. 4, E and F), and lung wet–to–dry weight ratios (fig. S6A). The repression of PARP-1 protein during VILI seen in wild-type mice was attenuated in miR-223–deficient mice (Fig. 4, G and H). In addition, analysis of PARP-1 enzyme activity revealed increased PARP-1 activity in miR-223−/y animals (Fig. 4, I and J). Moreover, miR-223–deficient mice also experienced increased tissue inflammation during pulmonary infection with methicillin-resistant Staphylococcus aureus (strain USA300). At early time points (4 hours after infection), miR-223−/y mice showed increased permeability of alveoli to proteins, elevated myeloperoxidase in the BAL, and augmented inflammatory cytokine production in pulmonary tissue (Fig. 5, A to G). In addition, mice lacking mmu–miR-223 showed significantly reduced survival during acute lung inflammation induced by S. aureus infection compared to wild-type mice (P < 0.05; Fig. 5H). Analysis of bacterial clearance revealed no significant difference between wild-type and miR-223−/y mice (Fig. 5, I and J), indicating that reduced survival in mice lacking mmu–miR-223 was unlikely to be due to reduced bacterial killing.

Fig. 4. Acute lung injury in miR-223−/y mice.

WT and miR-223−/y mice were exposed to VILI. (A) Albumin detected in mouse BAL fluid was measured in WT and miR-223−/y mice (means ± SEM; n = 5 mice in the WT control group and n = 7 mice in all others groups). (B) The neutrophil enzyme myeloperoxidase (MPO) was measured in BAL fluid from WT and miR-223−/y mice (means ± SEM; n = 5 mice in the WT VILI group and n = 6 mice in all others groups). (C and D) Interleukin-6 (IL-6) and Cxcl-1 protein concentrations in BAL fluid from WT and miR-223−/y mice were measured by enzyme-linked immunosorbent assay (ELISA) [means ± SEM; (C) n = 6 mice in the control group, n = 13 mice in the WT VILI group, and n = 12 mice in the miR-223−/y VILI group; (D) n = 6 mice in the control group, n = 8 mice in the WT VILI group, and n = 13 mice in the miR-223−/y VILI group]. (E) Representative slides of pulmonary histology for WT and miR-223−/y mice [hematoxylin and eosin (H&E) staining]. Scale bar, 100 μm. (F) Scoring of lung injury in slides from (E). (G and H) Pulmonary PARP-1 protein expression in WT or miR-223−/y mice at baseline (Ctr) or during acute lung injury (VILI) relative to the housekeeping gene β-actin (one representative blot of three is shown; quantification by densitometry, n = 3). (I) PARP-1 enzyme activity in WT or miR-223−/y mice (n = 5 mice in the miR-223−/y VILI group or n = 4 mice for all other groups). HPF, high-power field. (J) Representative immunohistochemical staining for PARP-1 activity in WT and miR-223−/y mice (*P < 0.05, ANOVA).

Fig. 5. Pulmonary bacterial infection in miR-223−/y mice.

Pulmonary infection in miR-223−/y mice (filled circles) and WT mice (empty circles) was induced by intratracheal administration of S. aureus. (A) Albumin detected in the BAL fluid was measured [(means ± SEM; n = 4 mice in the control group, n = 8 mice in the WT group infected with S. aureus, and n = 6 mice in the miR-223−/y group infected with S. aureus]. (B) Myeloperoxidase was measured in BAL fluid from miR-223−/y mice infected with S. aureus. IL-6 and Cxcl-1 proteins were measured in BAL fluid using ELISA: (C) IL-6 and (D) Cxcl-1 [(B and C) means ± SEM; n = 8 mice in the control group, n = 13 mice in the WT group infected with S. aureus, and n = 11 mice in the miR-223−/y group infected with S. aureus; (D) means ± SEM; n = 8 mice in the control group, n = 12 mice in the WT mouse group infected with S. aureus, n = 11 mice in the miR-223−/y group infected with S. aureus]. Transcripts were analyzed in mouse lung tissue for (E) IL-6, (F) IL-1β, and (G) Cxcl-1 by real-time RT-PCR relative to the housekeeping gene β-actin. [(E) means ± SEM; n = 4 WT or miR-223−/y mice in the control groups and n = 7 WT or miR-223−/y mice in the S. aureus infection groups; (F and G) means ± SEM; n = 4 WT or miR-223−/y mice in the control groups, n = 8 mice in the WT S. aureus infection groups, and n = 7 in the miR-223−/y S. aureus infection group]. (H) Survival of WT and miR-223−/y mice after pulmonary infection with S. aureus bacteria (n = 9 mice per group). Number of bacterial colony-forming units (CFU) 4 hours after infection with intratracheal S. aureus (I) (means ± SEM; n = 8 mice in the WT group and n = 7 mice in the miR-223−/y group) and in BAL fluid when mice were moribund (J) (means ± SEM; n = 7 mice in the WT group and n = 9 mice in the miR-223−/y group) (*P < 0.05 and §P < 0.05 in the post hoc analysis between WT mice and miR-223−/y mice after S. aureus exposure).

To demonstrate that increased lung inflammation and more severe acute lung injury in miR-223−/y mice were related to the miR-223 target gene PARP-1, we attempted to reverse the phenotype of miR-223−/y mice by repressing PARP-1 genetically using lentiviral transduction with a short hairpin RNA (shRNA) vector that was specific for PARP-1. Intratracheal administration of lentiviral shRNA was associated with reduced PARP-1 protein and PARP-1 enzyme activity at baseline and after VILI (Fig. 6, A to D). Functional studies analyzing the outcome after induction of VILI demonstrated that lentiviral shRNA–mediated PARP-1 repression in miR-223−/y mice was associated with a reversal of the lung inflammation phenotype (Fig. 6, E to K). Measurements of the alveolar epithelial injury markers RAGE (receptor for advanced glycation end-products) and the alveolar cell protein T1α revealed lower concentrations of both molecules in miR-223−/y mice with shRNA-mediated PARP-1 repression (fig. S7, A and B). In addition, these mice also showed lower numbers of TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling)–positive cells, implying a reduced cell death rate after PARP-1 repression (fig. S7C).

Fig. 6. Genetic knockdown of PARP-1 with shRNA in miR-223−/y mice with acute lung injury.

(A and B) Pulmonary PARP-1 protein knockdown in miR-223−/y mice using PARP-1 shRNA [one representative blot of three is shown; quantification by densitometry in (B), means ± SEM; n = 3 mice]. (C) Immunohistochemical staining for pulmonary PARP-1 activity in miR-223−/y mice treated with Parp-1 shRNA or control shRNA (n = 3 mice). (D) Quantification of immunohistochemical staining in (C) [means ± SEM; n = 6 mice in the nonventilated control shRNA group, n = 4 in the nonventilated Parp-1 shRNA group, and n = 10 or 5 mice in the control shRNA or Parp-1 shRNA ventilated (VILI) groups]. (E) Albumin detected in BAL fluid in miR-223−/y mice treated with control or Parp-1 shRNA and cytokine concentrations in BAL fluid measured by ELISA. (F) IL-6 and (G) Cxcl-1 protein. Pulmonary transcripts in miR-223−/y mice treated with control or Parp-1 shRNA for (H) IL-6, (I) IL-1β, (J) Cxcl-1, and (K) tumor necrosis factor–α (Tnfα), assessed by RT-PCR relative to the housekeeping β-actin gene [(F to K) means ± SEM; n = 4 mice in the control groups and n = 5 mice in the VILI groups] (*P < 0.05, ANOVA).

Collectively, these studies demonstrate that genetic deletion of miR-223 promoted increased lung injury during acute pulmonary inflammation induced by mechanical ventilation or bacterial infection. In addition, pulmonary edema and lung inflammation in miR-223−/y mice were reduced by overexpression of mmu–miR-223 and repression of PARP-1.

Inducing pulmonary miR-223 overexpression with nanoparticles dampens acute lung inflammation

We next pursued the possibility that overexpression of miR-223 was associated with lung protection during acute lung injury induced by mechanical ventilation. For this purpose, we treated wild-type mice endotracheally with nanoparticles containing synthetic miR-223 mimics (26). This treatment was associated with robust repression of PARP-1 transcripts in mouse alveolar epithelial cells (Fig. 7A). Previous studies using a PARP-1 inhibitor (PJ34) revealed that PARP-1 repression was beneficial during pulmonary inflammation, suggesting that miR-223–elicited PARP-1 repression could be useful for treating acute lung injury (23). Functional studies of lung inflammation induced by mechanical ventilation demonstrated that pulmonary overexpression of miR-223 in mouse lungs was associated with decreased albumin detection in the BAL fluid (Fig. 7B) and attenuated lung wet–to–dry weight ratios, indicating improved alveolar-capillary barrier function (Fig. 7C). Pulmonary overexpression of miR-223 was also associated with attenuated lung inflammation (Fig. 7, D to J). To address the effects of miR-223 overexpression in an infectious disease model of acute lung injury, we exposed mice intratracheally to methicillin-resistant S. aureus (strain USA300). Pulmonary overexpression of miR-223 attenuated lung edema, inflammation, and tissue injury during pulmonary infection with S. aureus (Fig. 8, A to G). Moreover, pulmonary miR-223 overexpression provided significantly improved mouse survival during acute lung inflammation induced by pulmonary S. aureus infection compared to mice treated with a control vector (P < 0.05, Mantel-Cox test; Fig. 8H). Analysis of pulmonary bacterial clearance revealed no difference between miR-223–overexpressing mice and control animals (Fig. 8, I and J). Pulmonary overexpression of miR-223 was associated with reduced PARP-1 protein expression (fig. S8, A and B) and a reduction in epithelial cell injury as measured by lowered T1α protein and reduced cell death (analyzed by TUNEL staining) (fig. S8, C and D).

Fig. 7. Pulmonary overexpression of miR-223 during acute lung injury in WT mice.

Nanoparticle-mediated overexpression of mmu–miR-223 in C57BL/6J mice before acute lung injury induced by ventilation (VILI). (A) Parp-1 transcripts measured by real-time RT-PCR relative to the housekeeping gene β-actin in isolated murine AT-II cells before and after nanoparticle-mediated delivery of miR-223 mimic (either mimic-control or mimic-223) (means ± SEM; n = 5 mice per group). (B) Albumin detection in murine BAL fluid was assessed by ELISA in nanoparticle-treated WT mice (nanoparticles contained either miR-223 mimic or control mimic; means ± SEM; n = 7 mice in the control groups and n = 11 mice in the acute lung injury groups). (C) Lung wet–to–dry weight ratio of nanoparticle-treated murine lungs at baseline and after acute lung injury induced by ventilation (VILI) (means ± SEM; n = 4 mice per group). (D and E) IL-6 and Cxcl-1 protein concentrations in murine BAL fluid were assessed by ELISA [(D) means ± SEM; n = 5 mice in the control groups and n = 9 mice in the acute lung injury VILI groups; (E) means ± SEM; n = 4 mice in the control groups and n = 9 or 10 mice in the mimic-223– or mimic-control–treated VILI groups]. (F) Representative slides of mouse pulmonary histology before and after treatment with nanoparticles containing miR-223 mimic (mimic-223) or control mimic (H&E staining). Scale bar, 100 μm. (G) Scoring of lung injury in histological sections in (F). (H to J) Pulmonary transcripts were assessed by real-time RT-PCR relative to the housekeeping gene β-actin for (H) IL-6, (I) Cxcl1, and (J) Tnfα [means ± SEM; n = 4 mice in the mimic-control–treated groups and n = 5 mice in the mimic-223–treated groups for (H); n = 4 mice in the mimic-control–treated groups and n = 3 and 5 mice in the mimic-223–treated groups for (I); n = 4 and 5 mice in the mimic-control–treated groups and n = 3 and 6 mice in the mimic-223–treated groups for (J)] (*P < 0.05 and §P < 0.05 in the post hoc analysis between “mimic-ctr”–treated and “mimic-223”–treated mice after VILI, Student’s t test or ANOVA).

Fig. 8. Pulmonary bacterial infection in C57BL/6J mice overexpressing miR-223.

After treating C57BL/6J mice with nanoparticles that contained either mimic-223 or mimic-control, mouse lungs were infected with S. aureus intratracheally. (A) Albumin detected in BAL fluid after S. aureus infection and nanoparticle treatment was measured by ELISA (means ± SEM; n = 5 mice per group). (B) Myeloperoxidase was measured in BAL fluid after S. aureus infection and nanoparticle treatment by ELISA (means ± SEM; n = 7 and 5 mice in the control groups and n = 10 mice in the S. aureus infection groups). Protein concentrations of (C) IL-6 and (D) Cxcl-1 were assessed by ELISA in control and S. aureus–infected C57BL/6J mice [(C and D) means ± SEM; n = 5 mice in the control groups and n = 10 mice in the S. aureus infection groups]. Mouse lung tissue was analyzed for (E) IL-6 and (F) IL-1β transcripts by real-time RT-PCR relative to the housekeeping gene β-actin [means ± SEM; (E) n = 9 mice in the mimic-control–treated group and n = 10 mice in the mimic-223–treated group; (F) n = 10 mice per group]. (G) Representative slides showing pulmonary histology for S. aureus–infected WT mice overexpressing miR-223 (mimic-223) or control mimic (H&E staining). Scale bar, 100 μm (means ± SEM; n = 5). (H) Mouse survival after pulmonary infection with S. aureus with or without treatment with nanoparticles containing miR-223 mimic or control mimic (n = 15 mice per group). (I) Lung bacterial colony-forming units 4 hours after infection of mice with intratracheal S. aureus and treated with nanoparticles containing miR-223 mimic or control mimic (means ± SEM; n = 5) (*P < 0.05, Student’s t test, ANOVA, or Mantel-Cox test).

Together, these data provide evidence that overexpression of miR-223 was associated with protection from lung inflammation and pulmonary edema during acute lung injury induced by mechanical ventilation or pulmonary S. aureus infection.

DISCUSSION

Here, we pursued the hypothesis that during acute lung injury, miRNAs are shuttled from PMNs to pulmonary epithelial cells, thereby providing a feedback mechanism to dampen inflammatory gene expression and lung inflammation. An experiment where human PMNs were coincubated with HPAEpiC revealed a selective increase in miR-223 expression by alveolar lung epithelial cells. Basal expression of miR-223 in HPAEpiC was very low, and external stimulation with LPS or fMLP failed to induce endogenous up-regulation of miR-223. Genetic studies using a coculture system comprising PMNs derived from miR-223−/y mice demonstrated that this increase in miR-223 reflected transfer of miR-223 from PMNs to the pulmonary epithelial cells. Moreover, further studies demonstrated that the release of miR-223 into the supernatant was increased upon activation of the PMNs and that transfer of miR-223 into alveolar epithelial cells was mediated by microvesicles. Studies of miR-223 target genes in human pulmonary epithelial cells indicated that anti-inflammatory functions of miR-223 during acute lung injury involved the miR-223 target gene PARP-1. Functional studies demonstrated that miR-223−/y mice were prone to acute lung inflammation induced by mechanical ventilation and that endotracheal application of miR-223–containing nanoparticles reversed this phenotype. Treatment with miR-223–containing nanoparticles provided protection from lung inflammation and pulmonary edema in wild-type mice during acute lung injury induced by mechanical ventilation or pulmonary infection with S. aureus. Together, these findings implicate PMN-dependent miR-223 transfer in an endogenous feedback mechanism that may limit excessive lung inflammation during acute lung injury via repression of PARP-1 (Figs. 7 and 8).

Previous studies have implicated miR-223 in PMN development and function. Human granulocytic differentiation is controlled by a regulatory circuitry involving miR-223, indicating that miR-223 may play a crucial role during granulopoiesis (27, 28). Other studies suggest that miR-223 expression might be driven by myeloid transcription factors (29). Meanwhile, studies using a loss-of-function allele in mice show that miR-223 negatively regulates myeloid progenitor cell proliferation and granulocyte differentiation and activation. In those initial studies, stimulation of mouse wild-type and miR-223−/y neutrophils with different concentrations of phorbol 12-myristate 13-acetate (PMA) led to production of lower amounts of reactive oxygen metabolites. Additionally, the observed percentage of wild-type versus miR-223−/y PMNs that underwent a respiratory burst was much lower. On the basis of those initial studies generating and characterizing the miR-223−/y mouse phenotype, the lack of miR-223 in the mutant mouse neutrophils induced a hypersensitive state in response to activating stimuli (25). As a consequence of this neutrophil hyperactivity, miR-223−/y mice spontaneously develop inflammatory lung pathology and exhibit exaggerated tissue destruction after endotoxin challenge (25). Our study indicates that miR-223 dampens lung injury induced by mechanical ventilation or by pulmonary infection with S. aureus.

We identified PARP-1 as a target gene of miR-223 in pulmonary epithelia. PARP-1 is a member of the PARP family, which is composed of enzymes that catalyze a posttranslational modification called poly(ADP-ribosyl)ation (30). PARP-1 is expressed in all mammalian cells except mature granulocytes, which is in line with our findings that miR-223–dependent lung protection involves the transfer of miR-223 to another cell type. PARP-1 is involved in the regulation of various biological activities, including DNA repair, DNA replication, transcription, and proteasomal degradation. However, PARP-1 has also been implicated in inflammation and in ischemia-reperfusion tissue injury (3032). Consequently, pharmacological inhibition or genetic deletion of PARP-1 has been shown to protect against inflammatory diseases, such as acute lung injury. Our findings of miR-223–dependent PARP-1 repression during acute lung injury are in line with an anti-inflammatory role for miR-223 shuttled by PMNs during pulmonary inflammation (31). Our observations in miR-223−/y mice using PARP-1 shRNA to inhibit PARP-1 expression during pulmonary inflammation are consistent with previous studies where PARP-1 inhibitors decreased lung permeability in lung injury mouse models (31). The improved barrier function we observed during induction of lung inflammation by mechanical ventilation in miR-223−/y mice with shRNA-mediated PARP-1 inhibition, indicated by reduced albumin leakage and diminished lung edema, is in agreement with observations in other inflammatory diseases. For example, pharmacological PARP-1 inhibition resulted in a reduction in proinflammatory cytokine production and in intestinal permeability in a murine model of experimental colitis (33). Similarly, our observations of diminished lung epithelial cell inflammation after transfer of miR-223 from neutrophils to alveolar epithelial cells indicate an anti-inflammatory and protective role for miR-223–mediated PARP-1 inhibition during acute lung injury.

Studies from the cancer field indicate that miRNAs can be released from microvesicles produced by cancer cells and implicate miRNA transfer in tumor expansion and metastasis. Our study suggests that inflammatory cells, such as PMNs, may release miRNAs in microvesicles to provide anti-inflammatory signals to pulmonary epithelial cells. Previous studies have shown that neutrophils can release anti-inflammatory microvesicles upon activation (34, 35). Lung infiltration by activated PMNs has been implicated as a key event in the development of lung inflammation during acute lung injury. The release of miR-223–containing microvesicles by PMNs may be a signal for the resolution of epithelial inflammation. Our findings suggest that neutrophils, often thought to only accelerate inflammation and injury, may play a critical role in injury repair potentially by transferring miR-223 to lung alveolar epithelial cells. These findings are consistent with previous studies implicating neutrophils in epithelial repair of the injured lung (36). Although they are crucial for fighting infection, neutrophils could also contribute to lung injury through several mechanisms that cause collateral damage (37). For VILI, neutrophil infiltration is a key event (38). High-pressure ventilation-induced pulmonary inflammation results in damage to the alveolar-capillary barrier, as shown in our study by increased albumin in the BAL fluid. During VILI, there is an increase in myeloperoxidase produced by neutrophils, indicating a PMN influx into the lungs. Exaggerated neutrophil activation and the release of microbicidal compounds and proinflammatory cytokines into the extracellular space can damage host tissues. Several studies have shown that depletion of granulocytes during lung injury attenuates inflammation and reduces the severity of lung injury (39, 40). In contrast, in other inflammatory disease models, there is also evidence that neutrophil depletion can aggravate inflammatory disease progression (41). Our study suggests that the infiltrating neutrophils provide an additional protective mechanism through transfer of miR-223 to the lung epithelium.

The phenomenon of neutrophil-epithelial cross-talk highlighted here was found to occur in different types of pulmonary epithelial cells. This is in line with previous observations that PMNs can migrate across connective tissue and epithelial barriers throughout the respiratory tract. We found some indication that miR-223 transfer primarily involves AT-II cells of the lung alveoli. AT-II cells have been described as relatively resistant to cell death during acute tissue injury, and miR-223–mediated AT-II cell protection may play a functional role during the acute phase of lung injury and repair.

Although our data suggest a functional role for miR-223–dependent miRNA transfer during acute lung injury, there are a number of major limitations to our study. Although we observed a protective role for miR-223 in the acute phase of lung injury and lung infection, the functional role of miR-223 during chronic states of lung disease and the impact on lung injury resolution remain unknown. Much of our study was carried out in miR-223−/y mice with global deletion of miR-223; additional studies in mice with miR-223 deletion in different lung tissue compartments may have the potential to identify additional miR-223 functions and other target genes. We identified PARP-1 as an epithelial miR-223 target gene mediating the protective role of miR-223 during acute lung injury. Given that miRNAs usually regulate multiple target genes, it is likely that there are miR-223 gene targets in addition to PARP-1 that could be involved in protection against pulmonary inflammation. We only examined global miR-223 expression in BAL fluid from patients with ARDS, but future studies will need to examine miR-223 expression in subgroups of patients suffering from lung injury and at diverse stages of disease.

MATERIALS AND METHODS

Study design

We combined several approaches to investigate miRNA transfer from neutrophils to alveolar epithelial cells during acute lung injury in mice. First, we established a coculture system in vitro using PMNs and pulmonary epithelial cells of human or murine origin to study miRNA transfer. Two different murine models of acute lung injury were used to investigate the transfer of miR-223 during pulmonary inflammation: a mechanical VILI model and a bacterial infection model of acute murine pneumonia. For histology scoring, the researcher was blinded to the treatment group and sample identity. Numbers of replicates for experiments are specified in the corresponding figure legends. Two different genotypes of mice were used to confirm the critical role of miR-223: wild-type and miR-223−/y mice. To further explore the relevant myeloid cell type as a source of miR-223, bone marrow chimeric mice were generated, and depletion studies were conducted. For miR-223 target identification, microarray analysis and luciferase assays were used. As proof of principle, miR-223 was overexpressed using nanoparticle-packaged miR-223 mimics. Analysis of miR-223 in human samples was performed on BAL fluid collected from 55 patients with ARDS within the first 7 days after diagnosis. Demographic and clinical data such as APACHE II scores and primary etiology are shown in Table 1. BAL fluid samples derived from healthy individuals served as controls. All eight control subjects were between 18 and 55 years of age, were not taking any medications, and were lifelong nonsmokers. There were five males and three females.

Cell culture and treatments

Primary alveolar epithelial cells (HPAEpiC, ScienCell Research Laboratories), human airway epithelial cells [Calu-3, American Type Culture Collection (ATCC)], murine alveolar epithelial cells (MLE-12, gift from I. Douglas, Pulmonary Sciences and Critical Care Medicine, School of Medicine University of Colorado, USA; E-10, gift from D. C. Thompson, Department of Clinical Pharmacy, School of Pharmacy University of Colorado, USA), and human promyelocytic leukemia cells (HL-60, ATCC) were cultured according to the manufacturer’s instructions and in the appropriate media. During coculture, all cell types were incubated in Hanks’ buffer.

Coculture experiments

Coculture experiments were set up using Transwell plates with 0.4-μm pore size polycarbonate membrane inserts (Corning) according to the manufacturer’s instructions. Depending on the readout, cocultures were carried out in 6-, 12-, or 24-well inserts. Pulmonary epithelial cells (Calu-3, HPAEpiC, MLE-12, and E-10) were seeded and grown to confluence in the lower chamber before coculture. During coculture, appropriate numbers [a ratio of 1:2 for all experiments, except Fig. 1E (ratio, 1:3)] of either freshly isolated human PMNs (activated with fMLP), murine isolated PMNs [activated with PMA (0.1 μg/ml)], or neutrophil-like HL-60 cells [differentiated 4 days with dimethyl sulfoxide (DMSO)] were added to the upper chamber, and coculture was carried out for 4 to 6 hours. Under control conditions, epithelial cells were treated with fMLP (1 mM) or LPS (10 μg/ml) for 6 hours. At the end of each coculture experiment, PMN-containing inserts were removed, and cells in the lower chamber were washed once with 1× phosphate-buffered saline (PBS), followed by an immediate lysis with QIAzol for RNA analysis. In the case of coculture experiments, luciferase-transfected pulmonary epithelial cells were transfected 24 hours before coculture, inserts were removed after 6 hours of coculture, and transfected cells in the lower chamber were further cultured for another 48 hours before lysis and luciferase activity measurements. In studies analyzing the epithelial microvesicle uptake mechanism, human pulmonary alveolar epithelial cells (HPAEpiC) were pretreated with 50 μM monodansylcadaverine (MDC; Sigma-Aldrich), an inhibitor of clathrin-dependent endocytosis (42), 30 min before coculture.

Generation of overexpressing miR-223 cell line

Green fluorescent protein (GFP)–miR-223 precursor construct, built in a lentiviral vector backbone (purchased from SBI), was produced by cotransfecting the plasmid with pPACK packaging mix (SBI into 293T cells, ATCC) to generate lentiviral particles. Virus-containing supernatants were harvested according to the manufacturer’s instructions (SBI), and Calu-3 were transduced with either GFP–miR-223 lentivirus or GFP–control lentivirus according to the manufacturer’s instructions.

Binding of miR-223 to miR-223 target vector or 3′UTR of PARP-1

Cells (Calu-3 or lentiviral-transduced Calu–lenti-223 or Calu–lenti-ctr) were transfected with PARP-1 3′UTR reporter construct (luciferase vector containing the 3′UTR of PARP-1 downstream the luciferase gene; OriGene) or miR-223 target vector (vector containing a target sequence for miR-223 downstream the luciferase gene; SwitchGear) or with respective control vectors. Forty-eight hours later, cells were harvested, and luciferase activity was measured using the Dual-Luciferase Reporter kit (Promega) (for the PARP-1 3′UTR reporter construct) or the LightSwitch Luciferase Assay Kit (SwitchGear) (for the miR-223 target vector) according to the manufacturer’s instructions.

Isolation of human PMNs

The collection and use of human PMNs from healthy volunteers were approved by the Colorado Multiple Institutional Review Board (COMIRB), and before each blood draw, written informed consent from each individual was obtained. For isolation of human PMNs, human blood was obtained by venipuncture from healthy volunteers. The blood was anticoagulated with an ACD (acid-citrate-dextrose) buffer and added to 3% dextran solution for sedimentation. To separate cells, we carefully layered anticoagulated whole blood on top of a density gradient cell separation medium of Ficoll and sodium diatrizoate (Histopaque-1077, Sigma-Aldrich). After centrifugation at 2700 rpm for 30 min at room temperature without break, supernatant was removed, PMNs containing pellet (from this step onward, all steps were carried out at 4°C) were resuspended, and residual erythrocytes were removed with two washes of ice-cold RCLB (red cell lysis buffer) and then washed twice in Hanks’ buffer (−). During counting and until experiments were started, PMNs were kept in Hanks’ buffer (−) and at 4°C. For activation, PMNs were transferred from Hanks’ buffer (−) to Hanks’ buffer (+) supplemented with 1 μM fMLP and incubated at 37°C. The viability of the cells after isolation was greater than 97% using the trypan blue dye exclusion test.

Isolation and purification of murine PMNs

Murine PMNs were isolated from the bone marrow of C5BL/6 or miR-223−/y mice using the EasySep Mouse Neutrophil Enrichment Kit (STEMCELL Technologies) according to the manufacturer’s instructions. The negative selection method was chosen because it is considered to result in less activation (compared to positive selection methods) during the purification procedure. Briefly, mice were euthanized, and the bone marrow was harvested from femurs and tibias. Cells were centrifuged and incubated with a cocktail of biotinylated antibodies for negative selection, again centrifuged and then incubated with EasySep D magnetic particles. Subsequently, cells were placed in an EasySep Magnet to separate and finally decant the nonlabeled purified murine PMNs. All steps were carried out at 4°C to avoid any activation of cells before starting the desired experiment.

Isolation of murine AMɸ

Murine AMφ were isolated as previously described (43).

Isolation of pulmonary endothelial cells

Murine pulmonary endothelial cells were isolated as previously described (44).

Lung injury score

Murine lung histology was scored blinded according to lung injury scoring criteria published previously (45).

Isolation and culture of murine BMDMɸ

Murine BMDMφ were isolated and activated as previously described (46).

Microvesicle isolation from PMN supernatant/human BAL samples

PMN-derived microvesicles were isolated by sequential ultracentrifugation as described previously (34). Briefly, after culture of freshly isolated PMNs, supernatants were collected, and cell debris was removed by two consecutive centrifugations (4000g for 20 min at 4°C). Then, to collect the PMN-derived microvesicles, the supernatants were ultracentrifuged at 160,000g for 60 min at 4°C in a Type 70.1 Ti rotor (Beckman Instruments Inc.). The pelleted microvesicles were further processed depending on the experimental setup; either the pelleted microvesicles were directly lysed in QIAzol for RNA isolation using the RNeasy kit (Qiagen) following the manufacturer’s instructions provided or kept in PBS for other experiments.

Microvesicle analysis with flow cytometry

Microvesicles were obtained from murine lungs by performing BAL with PBS that had been filtered through a 0.2-μm filter to remove particulate debris. Lavage fluid was then centrifuged at 200g for 5 min to pellet cells. One hundred microliters of cell-free supernatant was removed and incubated on ice for 30 min with fluorescently conjugated antibodies directed against CD45 (clone 30-F11, BD Biosciences) and CD11b (clone M1/70, BD Biosciences) at concentrations of 1:100. Control isotype antibodies were added to paired aliquots at similar concentrations. Samples were diluted with 900 μl of filtered PBS and taken to flow cytometry where calibrated fluorescent microvesicle sizing beads (Megamix, Biocytex) were added at a concentration of 27,000/ml. Microvesicles from cultured human neutrophils were analyzed by adding antibodies directed against CD45 (clone TU116, BD Biosciences) and CD11b (clone ICRF44, BD Biosciences) to 100 μl of cell culture supernatant for 30 min. Specimens were diluted in 900 μl of filtered PBS, and Megamix beads were added. Flow cytometry for all experiments was performed on a Gallios 561 cytometer (Beckman Coulter) with a wide-angle forward scatter aperture. Forward and side scatter thresholds were set using calibrated microbeads. Microvesicles were enumerated by collecting events between 0.2 and 0.9 μm in size after excluding microbeads based on their fluorescence.

Quantification of NET release from activated neutrophils

NETs were quantified using fluorimetry as described previously (47). Briefly, freshly isolated neutrophils were cultured in 96-well microtiter plates and stimulated with fMLP at a concentration of 1 μM. For inhibition of secretion of NETs, PMNs were pretreated with NETosis inhibitor Cl-Amidine (1 mM for 15 min) (Cayman Chemical). SYTOX Green, a non–cell-permeant DNA binding dye (Thermo Fisher), was added to the cells at 5 μM concentration to detect extracellular DNA. The degree of the fluorescence of treated cells was obtained after subtracting baseline fluorescence of medium only, and the data were normalized as 100% NET formation to the stimulus, leading to the greatest degree of NETosis (stimulation with fMLP) to more readily assess the fold reduction. The plates were read in a fluorescence microplate reader with a filter setting of 485 nm (excitation)/525 nm (emission).

Patient samples

Collection and use of patient samples were approved by the appropriate institutional review board (IRB) of each institution, and COMIRB approval had been obtained. BAL fluid was collected from 55 patients with ARDS within the first 7 days after diagnosis. Demographic and clinical data APACHE II scores and primary etiology are shown in Table 1. BAL fluid samples derived from healthy individuals served as controls. All eight control subjects were between 18 and 55 years of age, were not taking any medications, and were lifelong nonsmokers. There were five males and three females.

Mice

Experimental protocols were approved by the IRB at the University of Colorado Denver, CO. They were in accordance with the protection of animals and the National Institutes of Health guidelines for use of live animals. miR-223−/y mice were a gift from V. Dawson (Johns Hopkins University School of Medicine, Baltimore, MD), and wild-type C57BL/6J mice were obtained from the Jackson Laboratory. The miR-223−/y mice were generated on a C57BL/6 background. We had received mating pairs of miR-223 knockout mice on the background of C57BL/6 (male, miR-223−/y; female, miR-223−/−) and wild-type C57BL/6 mice. Both miR-223 knockout mice and the C57BL/6 mice that served for wild-type control mice have been bred and held in the same specific pathogen-free barrier unit. For experiments, miR-223−/y and wild-type mice have been matched for age and gender accordingly.

Mouse strains

miR-223−/y mice were generated and first characterized in 2008 (25). The miR-223 locus is located on the X chromosome and is transcribed independently of any known genes. The miR-223−/y mice were generated on a C57BL/6 background. These mice are fertile, were born at normal Mendelian ratios, and do not show any gross abnormalities when maintained in a pathogen-free environment. At the age of 8 to 10 weeks, these mice do not display any pulmonary abnormalities. We received mating pairs of miR-223 knockout mice on a C57BL/6 background (male, miR-223−/y; female, miR-223−/− and wild-type C57BL/6 mice). Both miR-223 knockout mice and the C57BL/6 mice that served for wild-type control mice have been bred and held in the same specific pathogen-free barrier unit. For experiments, miR-223−/y and wild-type mice have been matched for age and gender accordingly.

Murine model for VILI

Murine model for in situ acute VILI was performed as described before (7, 2022). Briefly, acute lung injury was induced with 3 hours of pressure-controlled ventilation using high inspiratory pressure of 45 mbar (2- to 5-mbar positive end-expiratory pressure). For that purpose, animals were anesthetized with pentobarbital [70 mg/kg intraperitoneally (ip) for induction and 20 mg/kg per hour for maintenance] and maintained at stable body temperature using a thermal feedback controller (heating tables and rectal thermometer probes), and fluid was replaced with normal saline (0.05 ml/hour ip). After tracheotomy, a tracheal tube was connected to a mechanical ventilator (Siemens Servo 900C, with pediatric tubing). For all experiments, mice were ventilated in a pressure-controlled ventilation mode with 100% inspired oxygen.

Pulmonary bacterial infection

For S. aureus experiments, bacteria were grown, as described previously, and administered by 50 μl of intratracheal inocula at 2 × 108 to 5 × 108 colony-forming units per animal (48). S. aureus strains were grown at 37°C to mid-exponential phase OD600 (optical density at 600 nm) = 1 (~3 hours) in CCY medium [3% yeast extract, 2% Bacto Casamino acids, 2.3% sodium pyruvate, 0.63% Na2HPO4, and 0.041% KH2PO4 (pH 6.7)].

Bone marrow chimera

To address the miR-223 transfer during VILI in vivo, bone marrow mouse chimeras were generated as described previously (7, 49), in which bone marrow was ablated by radiation in wild-type C57BL/6J mice followed by reconstitution with bone marrow derived from miR-223−/y (gene-targeted for miR-223) and vice versa (miR-223−/y→wild type and wild type→miR-223−/y; fig. S4A). Briefly, bone marrow cells harvested from femurs and tibias of euthanized male mice at 8 to 10 weeks of age were transplanted into irradiated (total dose of 2 × 6 gray) recipient mice immediately after irradiation. To monitor transplantation efficiencies, parallel experiments using the same regimen were carried out transplanting CD45.2 bone marrow into irradiated CD45.1 mice, and chimerism was examined by flow cytometric analysis at 6 to 8 weeks after irradiation, correlating with the time of induction of VILI in the experimental chimeric mice (fig. S4B). For this purpose, blood was taken from transplanted recipients; red blood cells were lysed, centrifuged, and resuspended in 200 μl of PBS containing 1% bovine serum albumin (BSA) to determine the percentage of neutrophils expressing CD45.2 in these chimeras (CD45.2→CD45.1); cells were incubated with anti–mouse-CD45.1 antibody (clone A20, BioLegend), anti–mouse-CD45.2 antibody (clone 104, eBioscience), anti–mouse–major histocompatibility complex II antibody (clone M5/114.15.2, eBioscience), and anti–mouse-Ly6G antibody (clone 1A8, BD Biosciences), and analysis was carried out.

PMN depletion

To deplete neutrophils 24 hours before induction of VILI, mice were given a single dose (1 mg, ip) of Ly6G-specific antibody (1A8, BioXcell). Control stands for no antibody treatment. These mice received equal volumes of vehicle (PBS) at the same time point as the experimental mice received the antibody dose. Flow cytometric analysis demonstrated that depletion was successful (fig. S4, D and E).

Monocyte and macrophage depletion

To deplete monocytes and macrophages 24 hours before induction of VILI, mice were given a combined dose of 100 μl of clodronate liposomes intratracheally plus 200 μl intravenously (ClodronateLiposomes.com). Control mice received equal amounts of control liposomes without clodronate at the same time points.

Closed-chest pulmonary intravital microscopy

Intravital microscopy of the subpleural microvasculature of male wild-type mice was performed as previously described (50). The dye-labeled mimic was purchased from Dharmacon (miRIDIAN Mimic) and was labeled with Dy547, a dye with absorbance and emission maximum at 557 and 570 nm, respectively. The mimic-transfected and Calcein-stained HL-60 cells (9 × 106 cells in 200 μl of PBS) were injected intravenously. After injection and intravital microscopy surgical preparation, mice were ventilated with high tidal volumes. High tidal volume ventilation was continued while performing confocal intravital microscopy using a Nikon A1R microscope with a CFI 75 Apo long working distance 25× objective (numerical aperture, 1.1).

Labeling, intravenous injection of PMN-like cells (HL-60), and visualization by costaining

Neutrophil-like differentiated (4 days with DMSO treatment) HL-60 cells were labeled using the Qtracker 655 Labeling Kit (Life Technologies) following the manufacturer’s instructions. Thirty minutes before ventilation and induction of acute lung injury, 10 million labeled cells were injected retro-orbitally in a total volume of 100 μl per mouse. After induction of VILI, lungs were harvested and paraffin-embedded, and AT-II cells were stained with FITC-pro-SPC. Costaining was analyzed by fluorescence microscopy.

Nanoparticle treatment

To obtain pulmonary overexpression of miR-223, we used a nanoparticle-based approach to transport synthetic mmu–miR-223 into the alveolar space. Neutral lipid emulsion (NLE) consists of 1,2-dioleoyl-sn-glycero-3-phosphocholine, squalene oil, polysorbate 20, and an antioxidant that, in complex with synthetic miRNAs, form nanoparticles in the nanometer diameter range (26). Mice were twice (96 and 48 hours before VILI) given a dose of 20 μg of synthetic mmu–miR-223 (mimic-223; Dharmacon) or respective control mimic (mimic-ctr) formulated with NLE according to the manufacturer’s instructions (MaxSuppressor In Vivo RNA-LANCEr II, Bioo Scientific) by intratracheal injection.

In vivo PARP-1 inhibition

For genetic knockdown of PARP-1 in vivo, 50 μl of saline-formulated high-titer lentiviral shRNA specifically inhibiting PARP-1 (formulated high-titer lentiviral shRNA specifically inhibiting PARP-1, “PARP-1 shRNA”; Sigma-Aldrich) or control lentivirus (empty vector control plasmid DNA that contains no shRNA insert; MISSION pLKO.1-puro Empty Vector Control Plasmid DNA; Sigma-Aldrich) was administered 72 hours before VILI into male miR-223 knockout mice by an intratracheal injection.

BAL and lung tissue harvest

Mice were euthanized after 3 hours of ventilation or no ventilation (control), and lungs were lavaged with 1 ml of sterile PBS (for protein analysis, etc.). After lavage, lungs were perfused with 10 ml of ice-cold saline via the right ventricle to remove any remaining blood in the pulmonary circulation. Lungs were excised and, together with the BAL fluid samples, immediately snap-frozen and stored at −80°C until further processed.

Lung wet–to–dry weight ratio

The lung wet weight–to–dry weight ratios have been performed as described previously.

Microscopic image acquisition

All standard histological images displayed were captured using an Olympus BX51 system microscope supplemented with an Olympus DP72 microscopic digital camera. Images were taken at room temperature using an UPIanApo 10×/0.40 objective lens providing ×100 magnification. For aquisition, cellSens Dimension (version 1.41) from Olympus was used.

Transcriptional analysis

Total RNA or separated fractions of mRNA and miRNA were isolated from human epithelial cells, murine lung tissue, or murine AT-II cells using QIAzol reagent and RNeasy kit following the manufacturer’s instructions (Qiagen). Transcript levels were determined by real-time RT-PCR using specific primer sets (Qiagen). For normalization of expression levels, endogenous expression levels of RNU-6 or SNORD48 or spiked-in cel–miR-39 were used for miRNA studies, and endogenous expression levels of β-actin were used for normalization of all the cytokines (IL-6, IL-1β, CXCL-1, and TNFα), as well as for PARP-1 expression levels.

Isolation of murine alveolar epithelial cells

AT-II cells were purified as described previously (51). Briefly, after induction of VILI, mice were euthanized by pentobarbital overdose. Lungs were lavaged with 3 × 1 ml of sterile PBS and then perfused with ice-cold saline via the right ventricle. As previously described, 3 ml of prewarmed dispase (BD) was instilled, followed by 0.5 ml of a low–melting point agarose, and chest cavity was iced for 2 min to harden the agarose plug. Lungs were removed intact and incubated in 1 ml of dispase at 37°C for 15 min. Tissue was dissociated manually, and cells were filtered (70/40 μm; Thermo Fisher), followed by centrifugation. Pelleted cells were resuspended and labeled with CD45 MicroBeads (Miltenyi Biotec), followed by negative selection using magnetic sorting technology. CD45-negative cells were next treated with Fc receptor–blocking monoclonal antibody (eBioscience) and labeled with anti–epithelial cell adhesion molecule (EpCAM) biotin (eBioscience) and anti-biotin DyLight 633. EpCAM-positive cells were positively selected using streptavidin magnetic microbeads (Miltenyi). Successful purification of AT-II cells was confirmed by analyzing cells for EpCAM expression using flow cytometric analysis (fig. S4C).

Bronchoalveolar lavage

To obtain BAL fluid after 3 hours of ventilation or from control mice, the tracheal tube was disconnected from the mechanical ventilator, and the lungs were lavaged with 1.0 ml of PBS. All removed fluid was immediately snap-frozen and stored at −80°C for later measurement of the albumin concentration or myeloperoxidase content.

ELISA (albumin, myeloperoxidase, and cytokines) in BAL fluid

BAL fluid samples were thawed to 4°C, and specific protein concentrations in BAL fluid were determined using specific ELISA kits according to the instructions given by the manufacturer: myeloperoxidase (Hycult Biotech) and albumin (Bethyl Laboratories).

Immunoblotting experiments

For detection of PARP-1 protein in cell lysates or murine lung tissue, mouse polyclonal PARP-1 antibodies (C2-10 and B-10; Santa Cruz Biotechnology) were used. For protein loading control, stripped blots were reprobed for β-actin using a mouse monoclonal anti-human β-actin primary antibody (CP01, Calbiochem). Briefly, cell lysate or murine lung tissue–derived protein samples were solubilized in reducing Laemmli sample buffer and heated to 90°C for 5 min. Samples were resolved on a 10% polyacrylamide gel and transferred to nitrocellulose membranes (Trans-Blot Turbo, Bio-Rad). The membranes were blocked for 40 min at room temperature in tris-buffered saline–Tween 20 (TBS-T) supplemented with 2% BSA. The membranes were incubated in 1:1000 mouse polyclonal PARP-1-antibody at 4°C overnight, followed by three 10-min washes in TBS-T. Subsequently, membranes were incubated in 1:5000 goat anti-mouse horseradish peroxidase (ImmunoPure Antibody, Thermo Scientific). After another series of three 10-min washes in TBS-T, proteins were detected by enhanced chemiluminescence (SuperSignal West Femto Maximum Sensitivity Substrate, Thermo Scientific). Protein levels were quantified by densitometry using the ImageJ and Gimp software.

T1α Western blot

Murine BAL fluid was centrifuged at 100,000g for 2 hours at 4°C. Pellets were boiled in Laemmli buffer and subjected to Western blotting using an antibody directed against T1α (Abcam), as previously described and shown to correlate with alveolar epithelial injury.

RAGE ELISA

RAGE levels in the murine BAL fluid were measured by ELISA (R&D) according to the manufacturer’s instructions, as previously described and shown to correlate with epithelial injury.

TUNEL assay

TUNEL staining was performed on lung sections using the In Situ Cell Death Detection Kit (Roche) according to the manufacturer’s instructions.

PARP enzyme activity

PARP activity in tissues was measured by an immunohistochemical method assessing PARP activity using biotinylated nicotinamide adenine dinucleotide (NAD) (31). Briefly, cryosections (10 μm) were fixed in 95% ethanol for 10 min at −20°C and then rinsed in PBS. Sections were permeabilized with 1% Triton X-100 in 100 mM tris (pH 8.0) for 15 min at room temperature. Reaction mixture [10 mM MgCl2, 1 mM dithiothreitol, and 30 μM biotinylated NAD+ in 100 mM tris (pH 8.0)] was then applied to the sections for 30 min at 37°C. Sections incubated with biotinyl-NAD+–free reaction mixture were used as controls. After three washes in PBS, incorporated biotin was detected by incubation with a 1:200 dilution of each of parts A and B of Vectastain Elite ABC kit (Vector Labs) for 90 min at room temperature. After three 10-min washes in PBS, color was developed with nickel-DAB substrate. Sections were counterstained in Fast Green, dehydrated, and mounted in VectaMount.

Data analysis

All statistical analysis was carried out using GraphPad Prism 5.0. All data are expressed as means ± SEM, and P < 0.05 was considered statistically significant. For comparison between two groups, two-sample t tests were used. In cases of comparison among more than two groups, ANOVA with Bonferroni correction for multiple comparison adjustments was applied. For all fold change data obtained from real-time quantitative PCR, we took log2 transformation before performing any statistical analysis. The Mantel-Cox test was used for analysis of survival curves. Because the human data derived from BAL fluid were not normally distributed (confirmed with D’Agostino-Pearson omnibus test; GraphPad Prism 5.0), we report the data as median and 25 to 75% percentiles. The data were analyzed using a nonparametric analysis (Mann-Whitney test).

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/9/408/eaah5360/DC1

Fig. S1. miR-223 expression after coculture and baseline levels of miR-223 (Ct values).

Fig. S2. NETosis inhibition and neutrophil-epithelial miRNA transfer detection.

Fig. S3. PMN-derived microvesicles in murine BAL fluid and miR-223 in isolated murine pulmonary endothelial cells.

Fig. S4. Bone marrow chimera and neutrophil depletion.

Fig. S5. miR-223 in macrophages.

Fig. S6. Wet–to–dry weight ratios in miR-223−/y mice.

Fig. S7. Genetic knockdown of PARP-1 in miR-223−/y mice.

Fig. S8. Epithelial cell death in miR-223–overexpressing C57BL/6J mice.

Table S1. Significantly down-regulated genes in miR-223–overexpressing epithelial cells.

REFERENCES AND NOTES

Acknowledgments: We acknowledge G. P. Downey and M. K. Koeppen for scientific discussions; J. McClendon and E. McNamee for technical assistance; J. M. Poth, M. Hübner, and J. Rink for experimental assistance; P. Soehnlein and A. Schumski for technical advice; I. S. Douglas and D. C. Thompson for providing the MLE-12 and E-10 cell lines; and V. Dawson for providing the miR-223−/y mice. Funding: This work was supported by NIH grants R01-DK097075, R01-HL098294, POI-HL114457, R01-DK082509, R01-HL109233, R01-DK109574, R01-HL119837, and R01-HL133900 (to H.K.E.); NIH grants R01-HL070952 (to M.R.B.) and R01-HL114381 (to P.M.H.); NIH/National Heart, Lung, and Blood Institute Acute Lung Injury Specialized Centers of Clinically Oriented Research grant P50 HL074024 (to T.J.S.); NIH grants HL080396 and ES023384 (to C.M.E.) and HL130938 (to C.M.E. and W.J.J.); a grant from the German National Research Foundation [Deutsche Forschungsgemeinschaft (DFG), SFB815] (to K.Z.); a grant from the Crohn’s and Colitis Foundation of America and North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition (to J.C.M.); grants from the DFG and the American Heart Association (AHA) (to V.N.); and from the AHA (to E.T.C., C.M.E., and T.W.). Author contributions: V.N. designed and performed all experiments, wrote the paper, and compiled all original data. H.K.E. provided major funding and essential mentoring and the overall research direction, wrote the manuscript, and reviewed all the original data. Major experimental and technical support was provided by K.S.B. (real-time PCR and immunoblotting), E.P.S. (closed-chest pulmonary intravital microscopy), T.A.P. (isolation of AT-II epithelial cells), B.D. (bone marrow chimeric mouse experiments), T.W. (histology), A.A.F. (bacterial infection model), L.B. (microvesicle detection), J.C.M. and G.T.F. (PARP-1 enzyme activity assay), M.W.G. (microvesicle preparation), W.J.J. (microvesicle detection), C.M.E. (bacterial infection model), and R.L.Z. (RAGE and T1α measurements and TUNEL assay). Extensive review of the manuscript and critical contributions to the concept and direction of the research were provided by E.T.C., E.P.S., T.J.S., M.R.B., A.A.G., M.W.G., W.J.J., R.L.Z., C.M.E., E.L.B., D.H., M.M., S.K., K.Z., and P.M.H. K.S.B. reviewed the original data. Extensive review of statistical analysis for all original data was provided by C.C. All authors contributed to the interpretation of the data and approved the final version of the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Microarray data have been deposited at www.ncbi.nlm.nih.gov/geo/ (accession number GSE47625). Requests for miR-223−/y mice should be made to V. Dawson.
View Abstract

Navigate This Article