|Home | About | Journals | Submit | Contact Us | Français|
Rationale: Acute respiratory distress syndrome is refractory to pharmacological intervention. Inappropriate activation of alveolar neutrophils is believed to underpin this disease’s complex pathophysiology, yet these cells have been little studied.
Objectives: To examine the functional and transcriptional profiles of patient blood and alveolar neutrophils compared with healthy volunteer cells, and to define their sensitivity to phosphoinositide 3-kinase inhibition.
Methods: Twenty-three ventilated patients underwent bronchoalveolar lavage. Alveolar and blood neutrophil apoptosis, phagocytosis, and adhesion molecules were quantified by flow cytometry, and oxidase responses were quantified by chemiluminescence. Cytokine and transcriptional profiling were used in multiplex and GeneChip arrays.
Measurements and Main Results: Patient blood and alveolar neutrophils were distinct from healthy circulating cells, with increased CD11b and reduced CD62L expression, delayed constitutive apoptosis, and primed oxidase responses. Incubating control cells with disease bronchoalveolar lavage recapitulated the aberrant functional phenotype, and this could be reversed by phosphoinositide 3-kinase inhibitors. In contrast, the prosurvival phenotype of patient cells was resistant to phosphoinositide 3-kinase inhibition. RNA transcriptomic analysis revealed modified immune, cytoskeletal, and cell death pathways in patient cells, aligning closely to sepsis and burns datasets but not to phosphoinositide 3-kinase signatures.
Conclusions: Acute respiratory distress syndrome blood and alveolar neutrophils display a distinct primed prosurvival profile and transcriptional signature. The enhanced respiratory burst was phosphoinositide 3-kinase–dependent but delayed apoptosis and the altered transcriptional profile were not. These unexpected findings cast doubt over the utility of phosphoinositide 3-kinase inhibition in acute respiratory distress syndrome and highlight the importance of evaluating novel therapeutic strategies in patient-derived cells.
Accumulation of polymorphonuclear neutrophils (PMNs) in the lung microvasculature and interstitial and alveolar compartments is a key feature of acute respiratory distress syndrome (ARDS), and inappropriate accumulation and/or activation of alveolar PMNs is proposed to cause local injury by release of oxygen radicals, proteases and neutrophil extracellular traps. Because of the challenges inherent in isolating alveolar PMNs in ARDS, their functional activity and responses to pharmacological agents are largely unknown.
ARDS alveolar PMNs are phenotypically distinct from their circulating counterparts, and display delayed apoptosis but preserved oxidative burst, phagocytosis, and neutrophil extracellular trap responses. ARDS blood PMNs have an intermediate phenotype between ARDS alveolar PMNs and healthy circulating cells, with a transcriptome showing significant alterations in cell survival and inflammatory pathways. This work improves our understanding of PMN function in ARDS and highlights the importance of studying tissue- and disease-specific cell populations.
The acute respiratory distress syndrome (ARDS) is characterized by diffuse alveolar injury and immune cell infiltration, resulting in intractable hypoxemia (1). Despite the adoption of lung-protective ventilation, mortality remains high (2), and many survivors have long-term physical or neurocognitive sequelae, with fewer than 50% returning to work (2). Management remains largely supportive with optimization of ventilator parameters (3), judicious fluid balance, and treatment of underlying causes; no pharmacological interventions have proven beneficial.
Accumulation of polymorphonuclear neutrophils (PMNs) in the lung microvasculature, interstitial and alveolar compartments is a key feature of ARDS (4), and an association has been reported between intensity of alveolar neutrophil infiltration and disease severity (5). Inappropriate accumulation and/or activation of PMNs within the alveoli is proposed to cause unrestrained release of oxygen radicals, proteases, and neutrophil extracellular traps (NETs). Due to challenges inherent in isolating alveolar PMNs (alvPMNs), their functional activity in ARDS is largely unknown. Historically, mouse models have been used as surrogates for alvPMNs (6); however, rodent neutrophils differ markedly from their human counterparts.
Traditionally, PMNs have been viewed as a homogeneous population of short-lived cells with limited transcriptional capacity and a fixed functional repertoire. More recently, concepts of long-lived PMNs, retrograde transendothelial migration, and PMN plasticity have emerged (7–12). Given recent demonstrations that PMNs can modify their transcriptional profile following an inflammatory insult (13), genome-wide transcriptional analysis provides a powerful tool to identify novel targets relevant to altered PMN functions, and it has been successfully applied in asthma, pulmonary arterial hypertension, and ARDS (14); however, studies in ARDS are based on the analysis of total peripheral white blood cells rather than purified neutrophils.
Lung epithelial cells (15) synthesize granulocyte-macrophage colony-stimulating factor (GM-CSF), a cytokine essential for alveolar macrophage function (16, 17) and surfactant homeostasis (18, 19). Conversely, during inflammation, GM-CSF potentiates superoxide anion production (20), promotes PMN survival (21), and is detrimental in models of acute lung injury (22). PMN longevity increases dramatically in ARDS, and some studies have identified GM-CSF as a major prosurvival mediator (23). While the molecular mechanisms governing PMN lifespan in ARDS are incompletely understood, the cytoprotective effect of GM-CSF in PMNs in vitro is class I phosphoinositide 3-kinase (PI3K) dependent (21). PI3K inhibition prevents lung tissue edema and leukocyte recruitment in models of ARDS (24), and inhibition of PI3K-γ in a sepsis model reduces end-organ damage (25). Following the early exuberant proinflammatory response in ARDS, patients develop immune paresis, increasing susceptibility to nosocomial infections (26). Recent studies have demonstrated that inhibition of PI3K-δ may improve PMN responses during this phase (27). These observations have triggered considerable interest in the therapeutic use of PI3K inhibitors in ARDS.
To our knowledge, we present the first comprehensive characterization of purified ARDS blood PMNs (bloodPMNs) and alvPMNs and the first genome-wide transcriptome analysis of purified ARDS bloodPMNs. We show that ARDS alvPMNs are hypersegmented, with enhanced CD11b and reduced CD62L expression, and display delayed apoptosis but preserved oxidative burst, phagocytosis, and NET responses. ARDS bloodPMNs display an intermediate phenotype, with a transcriptome showing significant alterations in cell survival and inflammatory pathways but little overlap with the PI3K-dependent gene signature. This work improves understanding of PMN function in ARDS and reveals that apoptosis of ARDS neutrophils is resistant to PI3K inhibition. Together, these observations strengthen the case for modulating PMN function in ARDS, but they cast doubt over the utility of PI3K inhibitors in this condition. Some results of these studies have been reported previously in the form of an abstract (28).
All studies complied with the Declaration of Helsinki. Written informed consent was obtained from the legal surrogate of patients with ARDS (UK08/H0306/17). Paired blood samples were obtained simultaneously from age- and sex-matched healthy volunteers (HVs) (UK06/Q0108/281).
Patients fulfilling the Berlin criteria for ARDS (29) were recruited from mixed medical-surgical and neurosciences-trauma intensive care units in a U.K. teaching hospital. Exclusion criteria were age younger than 18 years, HIV-positive status, or informed assent could not be obtained. The median tidal volume in the patients with ARDS was 7.64 ml/kg predicted body weight (interquartile range, 6.94–8.64 ml/kg). Patients underwent venipuncture, fiberoptic bronchoscopy (FOB), and bronchoalveolar lavage (BAL) within 48 hours of diagnosis. Sterile isotonic saline (3×50 ml) was instilled into a subsegmental bronchus. Recovery averaged 90 ml (range, 20–120 ml) and did not differ between patients and HVs. Bronchoalveolar lavage fluid (BALF) was immediately filtered and placed on ice. Control BALF was collected from patients (n=10) undergoing elective FOB for indications unrelated to infection or ARDS.
alvPMNs and autologous bloodPMNs were isolated from patients alongside bloodPMNs from age- and sex-matched HVs. alvPMNs were purified by immunomagnetic negative selection (RoboSep kit; STEMCELL Technologies, Vancouver, BC, Canada) (30) to greater than 90% purity and greater than 98% viability. bloodPMNs were purified over discontinuous plasma Percoll gradients (28).
PMNs were classified by nuclear morphology, with the assessor blinded to sample origin. Mature PMNs displayed three or four nuclear lobes connected by heterochromatin filaments. Band PMNs had less condensed chromatin and incompletely segmented nuclei (31), while hypersegmented PMNs possessed at least five lobes.
HV and ARDS bloodPMNs and unprocessed BALF were resuspended in phosphate-buffered saline containing 5% bovine serum albumin and protease inhibitor cocktail (cOmplete Mini, ethylenediaminetetraacetic acid free; Roche Diagnostics, Indianapolis, IN). Samples were stained with CD62L-allophycocyanin (clone 559772; BD Pharmingen, San Jose, CA), CD11b-fluorescein isothiocyanate (clone IM0530; Beckman Coulter, Brea, CA), or isotype-matched controls. PMN apoptosis was assessed after 20 hours by flow cytometry using fluorescein isothiocyanate–labeled Annexin V/propidium iodide (BD Pharmingen).
Neutrophils (5×106/ml) were primed with tumor necrosis factor (TNF)-α (R&D Systems, Minneapolis, MN), GM-CSF, control BALF, or ARDS BALF at 37°C for 30 minutes. The oxidative burst in response to formylmethionylleucylphenylalanine (FMLP) (100 nM), zymosan, or heat-killed Streptococcus pneumoniae (serum opsonized, five to seven particles per PMN) was assessed by luminol-dependent chemiluminescence (30).
BALF and serum mediators were measured by ELISA (LPS; Kamiya Biomedical Co., Seattle, WA), surviving, and leukotriene B4 (R&D Systems) or using the V-PLEX Human Biomarker 40-Plex Kit and the Human MMP 3-Plex Ultra-Sensitive Kit (both from Meso Scale Discovery, Rockville, MD). Where stated, BALF samples were corrected to the total protein concentration (Pierce bicinchoninic acid protein assay; Thermo Scientific, Rockford, IL).
PMN phagocytic capacity was assessed using 1 mg/ml pHrodo Red Staphylococcus aureus Bioparticles (Life Technologies, Carlsbad, CA). Internalization was verified by live confocal imaging.
bloodPMNs and alvPMNs (1×106/ml) incubated with SYTOX Green (5 μM; Life Technologies) were seeded onto 96-well optical microplates (BD Biosciences, San Jose, CA). NET formation was quantitated by hourly fluorescence measurements and verified by fluorescence microscopy using rabbit antihistone H3 (Ab5103; Abcam, Cambridge, UK).
Freshly isolated PMNs (1×106/ml) were fixed (4% paraformaldehyde), permeabilized (0.5% Triton X-100), and stained with antineutrophil elastase (1:1,000; Santa Cruz Biotechnology, Dallas, TX) and rhodamine phalloidin (1:200; Invitrogen, Carlsbad, CA).
Genome-wide transcriptomic changes were assessed in paired bloodPMNs from 12 consecutively recruited patients with ARDS who were representative of the full patient cohort in terms of age, sex, and ARDS severity and causation, as well as from 12 HVs. Further studies were undertaken in the following groups of HV bloodPMNs (n=10 per group): (1) 0 hours, vehicle control; (2) 6 hours, vehicle control; (3) 6 hours, recombinant human GM-CSF (rhGM-CSF) (1 ng/ml); (4) 6 hours, pan-PI3K inhibitor ZSTK474 (10 μM); and (5) 6 hours, rhGM-CSF plus ZSTK474.
Complementary DNA (cDNA) prepared from 2.5 ng of RNA using the WT-Ovation Pico RNA Amplification System (NuGEN Technologies, San Carlos, CA) was fragmented and labeled using the FL-Ovation cDNA Biotin Module V2 (NuGEN). Labeled cDNA was hybridized onto GeneChip Human Genome U133 Plus 2.0 oligonucleotide arrays (Affymetrix, Santa Clara, CA). Raw data (see online supplement) were normalized using the robust multiarray average method (32) and quality checked in R/Bioconductor. A linear model was fitted to normalized data for each probe set, and a post hoc test (Fisher’s least significant difference) generated fold changes and P values. Probes were identified as significant if their fold change was greater than 1.5 and the P value was less than 0.05, and they were mapped to pathways using Ingenuity Pathway Analysis software (QIAGEN, Redwood City, CA). The NextBio analysis platform (Illumina, San Diego, CA) was used to compare our ARDS data with (preanalyzed) publicly available transcriptomic data.
For each dataset analyzed, an appropriate linear mixed model was fitted. When required, the data were logarithmically transformed to meet the assumptions of the analysis (i.e., normally distributed errors and homogeneity of variance). Correction for false discovery rates in the transcriptional and cytokine analysis was done according to the method of Benjamini and Hochberg (33). The analyses were conducted in SAS version 9.3 software (SAS Institute, Cary, NC). The results are presented as means±SEM of independent experiments, with P<0.05 considered statistically significant. Full details of the number of patients and HVs included in each assay are provided in Figure E1 in the online supplement.
Twenty-three mechanically ventilated patients fulfilling the 2011 Berlin definition for ARDS were recruited. Their clinical, demographic, and physiological characteristics are outlined in Table 1. Standardized ventilator strategies in accordance with the ARDS Network low tidal volume protocol were employed. At sample collection, 4 of 23 patients had severe ARDS (PaO2/FiO2 ratio, ≤100 mm Hg), 11 of 23 had moderate ARDS (PaO2/FiO2 ratio, 101–200 mm Hg), and 8 of 23 had mild ARDS (PaO2/FiO2 ratio, 201–300 mm Hg). Sepsis and pneumonia were the commonest precipitating insults. Thirteen of 23 patients survived to discharge. All patients underwent FOB within 48 hours of diagnosis. PMNs constituted 69.7±4.2% of the differential leukocyte count in ARDS BALF (6.5±3.2% in control BALF) (Table 1). PMN abundance in BALF did not correlate with initial ARDS severity, abnormalities in gas exchange, or BALF protein concentration (data not shown).
Comparison of purified HV bloodPMNs, ARDS bloodPMNs, and alvPMNs revealed striking differences in cell morphology. While HV bloodPMNs had few hypersegmented PMNs, alvARDS PMNs displayed abundant hypersegmented nuclei and cytoplasmic vacuolation (Figure 1A); hypersegmented PMNs were not identified in control BALF. Immature “band” PMNs were also more common in ARDS bloodPMNs and alvPMNs than in control BALF (Figure 1A).
PMN activation status was assessed by confocal imaging of F-actin and cell surface staining of CD62L (L-selectin) and CD11b (Mac1). Prominent circumferential F-actin fluorescence was observed in a substantial proportion of the ARDS bloodPMNs compared with HV bloodPMNs (Figure 1B). The profile of surface receptor expression (upregulation of CD11b and downregulation of CD62L [34, 35]) on both ARDS bloodPMNs and alvPMNs was consistent with a primed and/or activated phenotype (Figure 1C).
Consistent with a previous report (23), we demonstrate that, after 20-hour ex vivo incubation, ARDS alvPMNs and bloodPMNs demonstrated a significantly reduced number of apoptotic cells (28.5±19.2% and 42.7±23% [percentage of apoptosis ± SEM], respectively) compared with PMNs isolated from HV blood (69.2±12%) (Figure 2A). The magnitude of the survival response exhibited by ARDS PMNs was equivalent (28.6±10%) to the cytoprotective effect conferred by incubating HV bloodPMNs with a maximally effective concentration of rhGM-CSF (1 ng/ml) (data not shown).
We next compared the ability of ARDS alvPMNs and bloodPMNs to mount an oxidative burst in response to FMLP, opsonized zymosan, and S. pneumoniae (Figures 2B–2D) (36). In contrast to unprimed HV bloodPMNs, which displayed minimal reactive oxygen species (ROS) generation to FMLP (Figures 2B and 2C), alvPMNs and bloodPMNs from patients with ARDS displayed robust ROS generation to all three stimuli, which in certain individuals exceeded those of TNF-α–primed HV bloodPMNs (Figure 2D). These data indicate basal priming and preserved NADPH oxidase responses of ARDS alvPMNs and bloodPMNs, challenging the notion that inflammatory PMNs become “exhausted” at peripheral sites.
Previous investigators have identified a defect in the phagocytic and microbicidal activity of neutrophils from patients with ARDS (37). However, in our cohort, flow cytometry and confocal microscopy demonstrated that the capacity of ARDS alvPMNs and bloodPMNs to phagocytose pHrodo Red–labeled S. aureus was fully preserved (Figures 3A–3B). This assay, supported by live-cell imaging, is based on differential fluorescence of this bioparticle in an acidic environment, ensuring that only organisms within functional phagosomes are detected.
In addition to phagocytosis and the oxidative burst, PMNs deploy NETs to facilitate pathogen clearance. NETs are composed principally of a DNA scaffold decorated with antimicrobial granule proteins, which acts as a mesh to immobilize pathogens (Figure 3C). In response to phorbol myristate acetate (Figures 3C and 3D) or pyocyanin (data not shown), ARDS alvPMNs and bloodPMNs displayed a similar capacity for NET production to that of HV bloodPMNs. Collectively, our results demonstrate preservation of the antimicrobial functions of ARDS alvPMNs and bloodPMNs.
To address whether factors present in the serum or BALF in patients with ARDS could account for the primed and/or prosurvival PMN phenotype, a series of multiplex ELISAs and bioassays were used to characterize the cytokine and growth factor profiles (n=18). As shown in Figure 4 and Figure E2, a consistent profile of increased acute-phase markers (e.g., C-reactive protein, serum amyloid A) and inflammatory cytokines (e.g., TNF-α, thymus and activation-regulated chemokine, monocyte chemoattractant protein 1, IL-8, and IL-6) was observed in ARDS serum and BALF compared with HV samples. By contrast, when BALF samples were corrected for total protein concentration, only C-reactive protein, IL-6, and monocyte chemoattractant protein 1 levels were significantly higher in ARDS samples than in control samples (Figure E2). Of note, at the single time point sampled, GM-CSF was quantifiable in only 5 of 23 ARDS BALF samples (lower limit of quantification, 7.6 pg/ml).
RNA transcriptomic analysis comparing freshly isolated ARDS bloodPMNs with HV bloodPMNs revealed a total of 1,319 altered genes (using cutoffs of fold change >1.5 and P<0.05; top-ranked up- and downregulated transcripts are shown in Figure 5A; a full list of all 1,319 genes and their relative fold changes is provided in Table E3). Using the NextBio analysis platform (which recognized 1,282 of the 1,319 differentially expressed genes), we compared these changes with publicly available datasets, finding a striking similarity to data in leukocytes from patients with severe burns or sepsis (38, 39); not only was there a strong overlap in gene changes, but also the direction of change correlated almost completely (Figure 5B). Ingenuity Pathway Analysis revealed a significant increase in pathways associated with the immune response, cytoskeletal remodeling, and mucin production, as well as significant decreases in cell death and/or apoptosis pathways, consistent with the neutrophil phenotype observed (Figure E3). Of note, of the 1,319 observed transcript changes, only 216 were differentially expressed in the same direction compared with HV bloodPMNs treated ex vivo with GM-CSF (Figure E4). The data discussed in this publication have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (GEO) and are accessible through GEO series accession number GSE76293 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE76293).
To establish the significance of the alveolar inflammatory environment, we sought to induce a phenotypic switch using ARDS BALF. Incubating HV bloodPMNs with Iscove’s modified Dulbecco’s medium with 10% autologous serum containing ARDS BALF (50:50 vol/vol) reduced the extent of apoptosis observed ex vivo at 20 hours (37.4±21.7%; 50:50 vol/vol control BALF, 70.5±12.4%) (Figure 6A). Furthermore, treatment with ARDS BALF (50:50 vol/vol) for 30 minutes enhanced FMLP-induced ROS production in HV bloodPMNs (Figure 6B) to a level comparable to optimally TNF-α– or GM-CSF–primed HV bloodPMNs (not shown), while control BALF had little effect. Thus, ARDS BALF supernatant recapitulated the prosurvival, primed NADPH oxidase signature seen in ARDS blood/alvPMNs.
A key objective of this study was to define the sensitivity of inflammatory PMNs to PI3K inhibition, since this pathway is pivotal in neutrophil survival, priming and/or activation, and ROS production (30). First, we confirmed that a pan-PI3K inhibitor, ZSTK474 (10 μM) (40), and to a lesser extent the p38 mitogen-activated protein kinase inhibitor SB741445 (10 μM), blocked GM-CSF–induced PMN survival in HV bloodPMNs in vitro (Figure 7A). ZSTK474 also blocked the survival effect of ARDS BALF supernatant on HV bloodPMNs (Figure 7B). However, neither compound restored normal neutrophil apoptosis in ARDS bloodPMNs (Figure 7C), implying that the aberrant disease-associated neutrophil survival is either irreversible or operates through a PI3K-independent pathway. Given that even the delayed addition of ZSTK474 to GM-CSF–treated HV bloodPMNs retained effectiveness in overcoming the prosurvival effect of this cytokine (data not shown), the involvement of a PI3K-independent pathway seems most likely. This conclusion is supported by the minimal overlap we observed between the transcriptomal signatures seen in the ARDS bloodPMNs and those seen in HV bloodPMNs treated with ZSTK474 (Figure E5). In contrast, ROS production by PMNs from HVs and patients with ARDS was completely abrogated by ZSTK474 (Figure 7D).
Isoform-selective PI3K inhibitors have been proposed as antiinflammatory agents in diseases such as ARDS (41). Hence, it is important to study their efficacy in patient-derived cells. ARDS alvPMNs have been little studied due to the difficulty of obtaining these cells from acutely unwell patients. Using purified blood and alveolar neutrophils from 23 patients with ARDS, we demonstrate a stepwise change from HV bloodPMNs through ARDS bloodPMNs to ARDS alvPMNs. ARDS alvPMNs, and to a lesser extent ARDS bloodPMNs, were distinct from HV bloodPMNs, with hypersegmented nuclei, increased CD11b expression, prolonged survival, and primed NADPH oxidase responses. Surprisingly, while the respiratory burst remained fully sensitive to PI3K inhibition, the prosurvival phenotype was not reversed by this strategy.
Few previous studies have included assessment of the characteristics of paired circulating and postmigration inflammatory tissue neutrophils. The hypersegmented CD11bhigh/CD62Llow cells with enhanced oxidative capacity we identified in ARDSblood and ARDS alvPMNs are reminiscent of circulating neutrophils isolated following endotoxin challenge (42, 43). These latter cells were immunosuppressive, inhibiting T-cell proliferation by release of hydrogen peroxide at the neutrophil–T-cell interface. Increased nuclear segmentation and oxidative potential has also been observed in tumor-associated neutrophils (44), associated with increased antitumor activity. Prolonged survival of ARDS alvPMNs has been reported previously and attributed to GM-CSF and/or granulocyte colony-stimulating factor in BALF (23); however, in contrast to Matute-Bello and colleagues, we did not observe significantly elevated levels of these cytokines, perhaps related to disease heterogeneity and differences in sampling time. Delayed apoptosis has been measured in neutrophils recruited to skin chambers versus paired circulating neutrophils, but synovial fluid–derived PMNs from patients with rheumatoid arthritis exhibited normal apoptosis (45). These differences correlated with local IL-1β levels, but IL-1β in the ARDS BALF in our study was not significantly elevated. The variable functional capacity of neutrophils from different locations underscores the need to explore the efficacy of potential therapeutic agents in disease-relevant cell populations.
We observed a proinflammatory cytokine profile in the blood of patients with ARDS, including several established priming agents. In our present study, ARDS blood and ARDS alvPMNs were functionally primed, and such cells have previously been implicated in lung injury (46, 47). We previously demonstrated that the pulmonary capillary bed can trap and “de-prime” neutrophils and that this mechanism may fail in ARDS, augmenting the circulating pool of these potentially injurious cells (48). Additional priming signals may be imparted during vascular transmigration (49), and ARDS BALF also primed the oxidative burst of HV bloodPMNs. Thus, a number of different factors may contribute to the pooling of primed neutrophils within the alveolar environment in ARDS.
ARDS bloodPMNs, and in particular ARDS alvPMNs, survived longer than HV bloodPMNs during ex vivo culture. This prosurvival phenotype was recapitulated by incubating HV bloodPMNs with ARDS BALF, implying that the enhanced longevity of these cells results at least in part from local exposure to mediators. However, while the pan-PI3K inhibitor ZSTK474 did not reduce the lifespan of ARDS bloodPMNs, it did reverse the prosurvival effects of both GM-CSF and BALF on HV bloodPMNs in culture, suggesting that the complex cytokine environment in BALF is not the only factor conferring PI3K resistance. It is possible that the duration of exposure to prosurvival mediators in vivo before inhibitor exposure is relevant and that survival signals imparted during transmigration will likewise have been entrained before PI3K inhibition. Finally, hypoxia may impart additional signals that are also relatively PI3K resistant, and hypoxia-inducible factor–dependent signaling was upregulated (ranked 14th in the pathways changed in this setting; see Table E3). Together with the limited overlap between the ZSTK474 bloodPMN or GM-CSF bloodPMN transcriptomes and the ARDS bloodPMN signature, our results suggest that targeting of PI3K during ARDS, while suppressing the damaging ROS formation, would not enhance cell clearance via apoptotic pathways.
We further interrogated the activation state of ARDS bloodPMNs by undertaking the first reported transcriptomic analysis of purified peripheral bloodPMNs from patients with ARDS. Our data revealed remarkable overlap between the transcriptomic profile of ARDS bloodPMNs and those published for mixed leukocytes in burns (see Figure 4B) and sepsis cohorts (33, 36). The top five canonical pathways identified in the ARDS blood neutrophil gene signature were the glucocorticoid, IL-4, p38 mitogen-activated protein kinase, antigen presentation, and CDC52 pathways. These were also within the top five pathways identified in the previous burns and sepsis cohorts using mixed leukocytes (38, 39). This suggests that, despite their heterogeneity, there is a strong commonality in a range of acute severe inflammatory disorders. This also provides possible directions for novel therapeutic interventions aimed at, for example, the IL-4 receptor or p38 pathways.
In this study, we sought to characterize the functional and transcriptional profile of PMNs isolated from the blood and airways of patients with ARDS. Although our study captured only 23 patients at a single time point, our data add considerably to knowledge of alvPMN and bloodPMN function and signaling profiles in ARDS. They challenge data derived from both animal models and healthy cells, with a marked primed and prosurvival phenotype, the latter recalcitrant to PI3K inhibition. We conclude that intervention with a PI3K inhibitor in these patients is unlikely to be an effective therapeutic strategy, since it will impair PMN bactericidal function without facilitating inflammation resolution. Our findings highlight the importance of working with patient-derived cells, particularly for biomedical research into novel treatments for ARDS.
The authors are grateful to David Bloxham, senior scientist in the Department of Haematology, Cambridge University Hospitals; Keith Burling, National Institute for Health Research (NIHR) Cambridge Biomedical Research Centre (BRC) Core Biochemistry Facility; Simon McCallum, NIHR Cambridge BRC Core flow cytometry; Epistem (Manchester, UK) for the transcriptomic data; and the patients and staff of the John V. Farman Intensive Care Unit and Neurosciences and Trauma Critical Care Unit, Cambridge University Hospitals NHS Foundation Trust.
This work was funded by a noncommercial grant from GlaxoSmithKline, with additional support from The Wellcome Trust, Papworth Hospital, the British Lung Foundation, and the National Institute for Health Research Cambridge Biomedical Research Centre. D.M.L.S. holds a Gates Cambridge Scholarship. C.S. is in receipt of a Wellcome Trust Early Postdoctoral Research Fellowship for Clinician Scientists (WT101692MA).
Author Contributions: J.K.J., E.R.C., A.C., D.H., A.A., M.B., C.S., and E.M.H.: contributed to the concept and/or design of the study; J.K.J., A.A., D.M.L.S., J.H., G.B., M.L., and K.H.: contributed to the acquisition of the data, and all authors contributed to the analysis and interpretation; and J.K.J., A.C., and E.R.C.: drafted the manuscript. All authors critically revised the manuscript for intellectual content and approved the final version before submission.
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.201509-1818OC on April 11, 2016