TNF-induced death signaling triggers alveolar epithelial dysfunction in acute lung injury

Abstract

The ability of the alveolar epithelium to prevent and resolve pulmonary edema is a crucial determinant of morbidity and mortality in acute lung injury (ALI). Tumor necrosis factor (TNF) has been implicated in ALI pathogenesis, but the precise mechanisms remain undetermined. We evaluated the role of TNF signaling in pulmonary edema formation in a clinically-relevant mouse model of ALI induced by acid aspiration and investigated the effects of TNF p55 receptor deletion, caspase-8 inhibition, and alveolar macrophage depletion on alveolar epithelial function. We found TNF playing a central role in the development of pulmonary edema in ALI, through activation of p55-mediated death signaling, rather than previously well-characterized p55-mediated proinflammatory signaling. Acid aspiration produced pulmonary edema with significant alveolar epithelial dysfunction, as determined by alveolar fluid clearance (AFC) and intraalveolar levels of receptor for advanced glycation end-products. The impairment of AFC was strongly correlated with lung caspase-8 activation, which was localized to type 1 alveolar epithelial cells by flow cytometric analysis. p55-deficient mice displayed markedly attenuated injury, with improved AFC and reduced caspase-8 activity but no differences in downstream cytokine/chemokine production and neutrophil recruitment. Caspase-8 inhibition significantly improved AFC and oxygenation, while depletion of alveolar macrophages attenuated epithelial dysfunction with reduced TNF production and caspase-8 activity. These results provide in vivo evidence for a novel role for TNF p55 receptor-mediated caspase-8 signaling, without substantial apoptotic cell death, in triggering alveolar epithelial dysfunction and determining the early pathophysiology of ALI. Blockade of TNF-induced death signaling may provide an effective early-phase strategy for ALI.

INTRODUCTION

Acute lung injury (ALI) / acute respiratory distress syndrome (ARDS) is a major cause of morbidity and mortality in critical care, and no treatment exists beyond supportive therapies (1, 2). ALI is characterized firstly by a severe disruption of the alveolar-capillary barrier leading to life threatening pulmonary edema, and secondly by intense pulmonary inflammation involving recruitment of blood leukocytes. Whilst leukocytes further aggravate damage to the alveolar-capillary barrier (3-5), edema can develop in the absence of neutrophils (6, 7) and manifest within minutes to hours of an initiating insult, well before substantial leukocyte infiltration into the alveolar space (8, 9). The alveolar epithelium plays a central role in providing the greatest resistance to edema formation (10), and its disruption significantly promotes the progression of ALI through both increased barrier permeability and decreased alveolar fluid clearance (AFC). Alveolar epithelial dysfunction, as quantitatively assessed by a decreased AFC rate, significantly influences mortality of ARDS through diminished resolution of edema (11).

It has been shown that alveolar epithelial injury in ALI is associated with apoptosis of epithelial cells (12). Apoptosis proceeds through two main pathways – the intrinsic pathway which is mediated through release of cytochrome c from mitochondria leading to activation of caspase-9, and the extrinsic pathway which is initiated by the ligation of death receptors on the cell surface, such as Fas receptor (Fas) and tumor necrosis factor (TNF) receptor 1 (p55), resulting in activation of ‘death signaling’ involving caspase-8. A number of studies have identified mechanisms through which the Fas ligand (FasL)/Fas interaction mediates epithelial cell apoptosis during ALI (13-15), and in some reports TNF-induced apoptosis has been implicated in ALI, although its significance has not been fully elucidated (16). Importantly, the majority of these studies focus on the relatively late phases of ALI, and it is not certain if apoptosis has any direct pathophysiological impact on the early development of this syndrome. Moreover, only a very small number of actual dead cells have been observed in lung tissue, e.g. up to a maximum of 10% (which included many apoptotic neutrophils) of total lung cells reported in autopsied ARDS lungs (13). Such a low degree of epithelial cell loss does not seem to be enough to explain significant deterioration in lung function observed in these experimental and clinical studies. The potential influence of death signaling per se on the functional status of the alveolar epithelium, before actual cell death that is the final event in the apoptotic process takes place, has not been investigated.

In this study we sought to investigate the role of TNF, an early phase proinflammatory cytokine as well as a death ligand, in pulmonary edema formation during ALI. TNF has been often implicated in the pathogenesis of ALI (17), but therapies to block its actions have been applied to critical illness in a rather premature manner, without a proper understanding of the precise mechanisms through which it functions. Here we present important in vivo evidence that alveolar macrophage-derived TNF plays a crucial role in triggering alveolar epithelial dysfunction leading to pulmonary edema, through activation of its p55 receptor-mediated death signaling involving caspase-8, rather than previously well characterized proinflammatory signaling. Our findings reveal a novel concept that TNF-mediated death signaling per se, without producing significant apoptotic cell death, determines the functional derangement of the alveolar epithelium and early pathophysiology of ALI. This offers important insights into potential new treatments for ALI targeting death signaling.

MATERIALS AND METHODS

Model of experimental lung injury

All protocols were approved by the Ethical Review Board of Imperial College London and carried out under the authority of the UK Home Office in accordance with the Animals (Scientific Procedures) Act 1986, UK. We used male wildtype (WT) C57BL6 (Charles River) and TNF p55 receptor knockout (p55^−/−) mice (Jackson Laboratories). The surgical preparation has been described in detail previously (18). In brief, mice were anesthetized by intraperitoneal injection of ketamine (80mg/kg) and xylazine (8mg/kg), and underwent tracheostomy and non-injurious low tidal volume ventilation (8-9ml/kg tidal volume, 2.5cmH₂O positive end-expiratory pressure, 120 breaths/minute and FiO₂ of 1.0) using a custom-made mouse ventilator-pulmonary function testing system. The left carotid artery was cannulated for monitoring arterial blood pressure and blood gases, and for saline infusion (0.4 ml/h). Once instrumentation was complete, baseline blood gases, peak inspiratory pressure and respiratory mechanics were recorded. Subsequently, hydrochloric acid was instilled via the tracheostomy tube using a fine catheter, and ventilation continued for 90 or 180 minutes depending on endpoints measured. After pilot experiments titrating the dose and concentration, we were able to create a robust, reproducible lung injury over 3 hours using 50μl of 0.15M hydrochloric acid (pH 1.5).

Physiological measurements

Airway pressure, gas flow, and mean BP were continuously monitored throughout the experimental protocol, while respiratory mechanics and arterial blood gases were assessed at pre-determined intervals. Respiratory system elastance was measured using the end-inspiratory occlusion technique as described previously (19). Sustained inflation of 30cmH₂O for 5 seconds was performed every 30 minutes to maintain alveolar recruitment.

Lung edema and bronchoalveolar lavage fluid analyses

At the end of the 3 hour experimental protocol mice were sacrificed, and the left lung was removed, weighed and dried at 60°C for 24 hours for wet:dry weight analysis. Bronchoalveolar lavage of the right lung was performed with 400μl of saline as described previously (18), and the samples were centrifuged at 1500rpm. Protein levels in bronchoalveolar lavage fluid (BALF) were quantified by Bradford assay (Bio-Rad Laboratories), and BALF levels of IL-6, TNF, KC, and MIP-2 were determined using ELISA assay kits (R&D Systems).

Lung permeability analyses

In some animals, Alexa Fluor 594-conjugated albumin (Invitrogen) was injected via the external jugular vein at 30 minutes before the end of the 3-hour protocol. Alveolar-capillary barrier permeability index was calculated by the ratio of BALF:plasma fluorescence (Biotek) as described previously (20).

Lung leukocyte quantification

In a separate series of experiments, following termination, left lungs were processed for histological staining (see below) while right lungs were used to prepare lung single cell suspensions as previously described (4, 21). In brief, lungs were mechanically disrupted using a GentleMACS tissue dissociator (Miltenyi BioTec) for 1 minute in IC fixative (eBioscience). They were then sieved through a 40μm filter in a flow cytometry wash buffer (PBS with 2% fetal calf serum, 0.1% sodium azide and 5mM EDTA). Cells were stained in the dark for 30 minutes with fluorophore-conjugated anti-mouse antibodies to CD45 (clone 30-F11, Biolegend), CD11b (M1/70, Biolegend), NK1.1 (PK136, Biolegend), Gr-1 (RB6-8C5, Biolegend), and Ly-6C (AL-21, BD Biosciences). Samples were analyzed using a CyAn ADP flow cytometer (Beckman Coulter) after the addition of AccuCheck counting beads (Invitrogen). Analysis was performed using FlowJo software (Tree Star).

Histology

Left lungs of this group of animals underwent intratracheal instillation with 4% paraformaldehyde at a transpulmonary pressure of 15cmH₂O. They were subjected to paraffin embedding, and 5μm sections were attained. Sections were stained with hematoxylin and eosin (H&E; Sigma-Aldrich) for examination. In addition, TUNEL staining was carried out to detect apoptotic DNA damage using the TACS.XL diaminobenzamine (DAB) in situ apoptosis detection set (Trevigen) as per manufacturer’s instructions. Apoptosis index was quantified on 20 random sections (at x400 magnification) of the left lung per mouse by an investigator blinded to the groups (16). TUNEL positive cells were counted as those with a distinct DAB positive nucleus within the alveolar wall. Images were acquired using a BX-60 light microscope (Olympus) to a digital AxioCam camera (Zeiss) and with KS300 v3.0 software (Zeiss).

Measurement of alveolar fluid clearance

A third group of animals underwent acid aspiration for functional determination of alveolar fluid clearance (AFC). This was measured using an ex vivo, in situ set-up as previously described in detail (22). In brief, 90 minutes post acid instillation animals were sacrificed and exsanguinated. Mice then underwent intratracheal instillation with 700μl of an iso-osmolar medium containing 5% low-endotoxin bovine serum albumin (Sigma-Aldrich) and 50μg/ml fluorescent Alexa Fluor 594-conjugated BSA (Invitrogen), followed by removal of a first aliquot of 200μl (t₀ sample). Mice were placed on a continuous positive airway pressure system delivering 100% oxygen at 8cmH₂O, and maintained at a temperature of 36.5-38°C throughout. Thirty minutes after instillation, a surgical pneumothorax was induced to maximize recovery of the remaining instillate (t₃₀ sample) from the lungs. Percentage AFC over 30 minutes was determined by increase in fluorescence between the t₀ and t₃₀ sample. Soluble levels of the receptor for advanced glycation end-products (RAGE) were measured by ELISA (R&D Systems) in t₃₀ samples to determine relationship with AFC rate.

Measurement of caspase-8 activity

Caspase-8 activity was measured in lung homogenates using a Caspase-8/FLICE Fluorometric Assay Kit (Biovision) as per manufacturer’s instructions. The left lungs of animals (post AFC measurement) underwent a standardized homogenization in the cell lysis buffer provided, using a hand-held homogenizer for 30 seconds.

Localization of caspase-8 activation

In the fourth group of animals, localization of caspase-8 activation within the lung was evaluated in WT mice using flow cytometry. Single lung cell suspensions were prepared from lungs of uninjured and injured animals at 90 minutes after acid instillation. In order to efficiently recover epithelial and endothelial cells, a tissue digestion protocol with Dispase, a modification of the previously published methods (21, 23-25), was employed. Lungs were instilled with 1ml DMEM containing 1mg/ml sterile filtered Dispase (Invitrogen) via the tracheostomy, and placed in the same Dispase solution for 30 minutes at room temperature. Lungs were subsequently placed in ice-cold sterile DMEM/2.5% HEPES with 0.01% DNase (Roche), and the parenchymal tissue was separated away from the bronchial tree and gently minced. This suspension of the distal lung tissues was passed through a 40μm filter, washed and reconstituted with ice-cold DMEM/2.5% HEPES. Cells were incubated with 3.33ul of FAM-IETD-FMK caspase-8 reagent (Immunochemistry) for 60 minutes at 37°C in the dark, as per manufacturer’s instructions, to enable this fluorochrome-conjugated caspase-8 inhibitor based compound to permeate the cells. The reagent binds specifically to activated intracellular caspase-8, which can be later detected in individual cells by flow cytometry (in the FITC channel). Cells were then washed, resuspended in DMEM/2.5% HEPES and incubated for 15 minutes at 37°C allowing any unbound caspase-8 reagent to diffuse out of the cells. After a further wash, cells were stained with fluorophore-conjugated anti-mouse antibodies to pan-endothelial marker - CD31 (clone MEC 13.3, Biolegend); the pan-leukocyte marker - CD45; the pan-epithelial marker - epithelial cell adhesion molecule (EPCAM) (G8.8, eBioscience); and the type 1 alveolar epithelial cell (AEC) marker – type 1 cell alpha protein (T1alpha) (8.1.1, eBioscience), and analyzed by flow cytometry as described above. Of note, lung cell suspensions were unfixed and kept on ice throughout the protocol, except during the incubation/wash steps for the FAM-IETD-FMK caspase-8 reagent binding.

In vivo caspase-8 inhibition

In the fifth series of experiments, 4mg/kg of the selective caspase-8 inhibitor Z-VAD-IETD (BD Biosciences), or DMSO as the vehicle, was intravenously administered (via the external jugular) to WT mice 5 minutes before intratracheal acid instillation, and alveolar fluid clearance and oxygenation were measured at 90 minutes.

Depletion of resident alveolar macrophages

In the final series of experiments, WT mice were anesthetized 48 hours prior to acid aspiration and underwent laryngoscopy as previously described (22) for intratracheal instillation of 75μl of clodronate or PBS encapsulated into liposomes (Encapsula NanoSciences), to deplete resident alveolar macrophages. We used a previously described flow cytometric identification of resident alveolar macrophages to evaluate the extent of their depletion (9). In brief, lung cell homogenates were prepared as described above and stained in the dark for 30 minutes with fluorophore-conjugated anti-mouse antibodies to CD45, F4/80 (CI:A3-1, Biolegend), CD11b, and CD11c (N418, eBioscience). Alveolar macrophages were identified as CD45^hiF4/80^hiCD11c^hi and CD11b^lo. AFC was determined as described above, and soluble levels of TNF, Fas ligand (FasL) and RAGE were measured by ELISA (R&D Systems) in t₃₀ samples.

Statistics

Data are expressed as means±SD or median±interquartile range (if non-parametric) and analyzed using SPSS version 20 (IBM). The model assumption of normality of residuals was assessed by QQ plot and Shapiro-Wilk test. Statistical analyses of data were made using either a 2-tailed student’s t-test or a 1-way ANOVA with Bonferroni tests for multiple comparisons, or Mann-Whitney/Kruskal-Wallis tests where data was identified as being non-parametric. Time-courses were analyzed using a t-test of final end-point values. We used Pearson’s correlation coefficients to test the relationships between continuous variables. Statistical significance was defined as p<0.05.

RESULTS

TNF p55 receptor plays a crucial role in the development of pulmonary edema in acute lung injury

Acid instillation induced an immediate ‘spike’ in respiratory system elastance (figure 1A) due entirely to the presence of fluid within the airways, because saline instillation induced the same initial elastance change, which rapidly returned to baseline as the fluid was absorbed. In acid instilled WT animals, after this initial spike, elastance remained high and started to increase further towards the end of the 3-hour protocol, indicating formation of pulmonary edema. In sharp contrast, p55^−/− mice displayed a steady recovery in elastance from the initial spike towards the baseline, in a similar fashion to saline-instilled WT animals. The p55^−/− strain had significantly improved arterial oxygenation compared to WT animals (figure 1B). Acid instillation in WT animals produced at 3 hours considerable increases in lung wet:dry weight ratio (figure 1C), total protein levels in BALF (figure 1D) and alveolar-capillary barrier permeability index (figure 1E), which were all markedly attenuated in p55^−/− mice. This effect of TNF p55 receptor signaling on alveolar edema development was apparent as early as 60-90 minutes after injury was initiated, by which point the elastance curves had clearly diverged.

Figure 1.

Mice deficient in the TNF p55 receptor display considerable physiological protection from acid-induced acute lung injury. Time course of A) respiratory system elastance (N=6 saline, N=10 WT, N=8 p55^−/−, P<0.0001 by t-test of end-point 3 hour values) and B) arterial oxygenation showing divergence between WT and p55^−/− strains (N=6 saline, N=12 WT, N=9 p55^−/−, P<0.0001 by t-test of end-point 3-hour values). C) Lung wet:dry weight ratio at 3 hours shows significantly reduced water content in p55^−/− mice (N=6 per group). Alveolar-capillary barrier integrity at 3 hours, as measured by D) bronchoalveolar lavage fluid (BALF) total protein concentrations (N=5-6 per group) and E) alveolar capillary barrier permeability index, show significantly improved barrier function in p55^−/− mice (N=4 per group). *** P<0.001; **P<0.01.

TNF-induced early development of pulmonary edema is not due to enhanced downstream proinflammatory responses

To investigate the role of TNF-induced downstream proinflammatory signaling in this model, we studied levels of cytokines/chemokines and leukocyte recruitment within the lung at 3 hours. We first confirmed a clear up-regulation of soluble TNF (ligand) itself in BALF both in the WT and p55^−/− strains at comparable levels (figure 2A). Acid aspiration led to significant increases in alveolar IL-6, KC and MIP-2 (figure 2B-D), but there were no differences between WT and p55^−/− strains.

Figure 2.

Levels of pulmonary inflammation in p55^−/− mice are high and similar to injured WT mice. Acid induced lung injury leads to an upregulation at 3 hours in A) TNF, B) IL-6, C) keratinocyte-derived chemokine (KC) and D) macrophage inflammatory protein (MIP)-2 in both WT and p55^−/− mice (N=8-10 per group). E) Neutrophils and monocytes were first gated in lung cell suspensions as CD45^hiCD11b^hiNK1.1^lo events (R1 + R2). Subsequently, neutrophils were identified as Ly-6C^interLy-6G^hi (R3), whereas inflammatory monocytes were identified as Ly-6C^hiLy-6G^lo (R4). We have previously shown that following the cell fixation procedure, the Gr-1 antibody used (clone RB6-8C5) only recognizes the Ly-6G epitope on cells (21). Thus events showing positive staining with this antibody here were defined as Ly-6G^hi, rather than Gr-1^hi. Flow cytometry of right lung homogenates obtained at 3 hours after acid aspiration shows a significant sequestration of neutrophils and inflammatory monocytes in all strains (N=5-7 per group). F) Hematoxylin & eosin sections of WT and p55^−/− animals at 3 hours after acid instillation. Acid instillation induces significant increases in alveolar wall thickening, intra-alveolar proteinaceous material, and alveolar edema in WT animals. These parameters representing alveolar-capillary barrier breakdown are reduced in p55^−/− animals whilst leukocyte infiltration is present in both strains (scale bar represents 50μm).

*** P< 0.001; ** P< 0.01.

Cytometric analysis showed minimal neutrophilia in BALF in all groups (% neutrophils in BALF cells: WT uninjured, undetectable: WT acid, 3.3±1.7%: p55^−/− acid, 4.1±2.4%). There were however significant increases in lung-sequestered neutrophils and inflammatory (Ly-6C⁺) monocytes (figure 2E) at 3 hours, again, with no differences between WT and p55^−/− mice. This was consistent with lung histology where differences in parameters related to barrier breakdown (e.g. intra-alveolar proteinaceous material, alveolar edema) were apparent between the two strains, whereas leukocyte infiltration seemed similar (figure 2F).

Acid-induced lung injury produces early activation of TNF p55 receptor-mediated death signaling

Based on the above findings, we hypothesized that TNF p55 receptor-mediated activation of death signaling, rather than proinflammatory signaling, may play an important role in the development of ALI. We examined lung sections for nuclear changes suggestive of late-stage cellular apoptosis (figure 3A-D). Acid instillation produced a subtle increase in the number of TUNEL positive cells within the alveoli in WT animals at 3 hours, which was reduced in p55^−/− mice (figure 3E). Despite detecting only minimal numbers of dead cells, we found that lung caspase-8 activity was considerably upregulated in WT animals, which was markedly attenuated in p55^−/− mice (figure 3F). Examination of the time course of caspase-8 activation in WT mice revealed that the activity was significantly increased even at 90 minutes (figure 3G). As 90 minutes was also the time at which physiological differences between WT and p55^−/− mice became apparent (figures 1A and B), subsequent measurements were determined for this time point.

Figure 3.

Attenuated alveolar death signaling in lungs of animals deficient in the TNF p55 receptor after acid aspiration. DAB TUNEL staining of lung sections at 3 hours post instillation (scale bar represents 10μm) - A) Uninjured WT mice showed undetectable TUNEL positive cells. B) Uninjured WT section treated with a nuclease enzyme to promote DNA strand breaks, illustrating the maximum number of alveolar nuclei which could be stained TUNEL positive (~80 per high power field). C) Occasional sections of acid instilled WT animals showed some TUNEL positive events (black arrowheads), as shown here. D) Sections of p55^−/− mice showed much lower numbers of TUNEL positive events. E) Apoptotic index, defined as average number of TUNEL positive nuclei per high power field based on the examination of at least 20 random sections, scored less than one in acid instilled WT animals (N=5 per group), suggesting minimal actual cell death despite substantial physiological dysfunction. This index was further reduced in p55^−/− mice. F) Despite minimal nuclear changes suggestive of end-stage apoptotic cell death there were significant increases in caspase-8 activity (at 3 hours) in WT injured animals which were substantially attenuated in p55^−/− mice (N=5 per group). G) The upregulation of caspase-8 activity is an early event in the pathogenesis of acid induced lung injury (N=5 per time point). *** P<0.001; * P<0.05.

Caspase-8 activation occurs predominantly in type 1 alveolar epithelial cells

To determine the localization of caspase-8 activation in this model, we developed a flow cytometry-based method to evaluate the levels of activated caspase-8 in lung cell suspensions, using a fluorochrome-conjugated cell-permeable caspase-8 specific inhibitor compound FAM-IETD-FMK. ‘Unfixed’ (to avoid loss of active caspase-8 associated with cell fixation procedures) lung single cell suspensions were prepared through dispase digestion as previously described (23, 24, 26), and caspase-8 activity was assessed in type 1 AECs, endothelial cells and leukocytes. Flow cytometric analysis indicated that the level of activated caspase-8 was predominantly localized to type 1 AECs (figure 4A). There was a 8-fold increase in median fluorescent intensity of FAM-signals in injured type 1 AECs compared to uninjured cells; in sharp contrast, no differences in fluorescent signals were observed between injured and uninjured animals in endothelial cells (figure 4B), and we found only negligible levels of activated caspase-8 in leukocyte populations (figure 4C).

Figure 4.

Acid induced lung injury induces activation of caspase-8 predominantly within type 1 alveolar epithelial cells (AEC). A) After exclusion of debris (using forward scatter and side scatter) type 1 AECs were identified as CD45^loCD31^loEpCAM^hiT1alpha^hi cells. This cell population showed a significant increase in caspase-8 activity during acid induced lung injury (solid histograms) compared to control uninjured animals (dashed histogram), as reflected by the difference in the median fluorescent intensity between the two groups (N=6). The activity of caspase-8 is localized predominantly to the lung epithelium which shows a log fold higher signal compared to B) lung endothelium (identified as CD45^loEpCAM^loT1alpha^loCD31^hi) and C) leukocytes (identified here as CD31^loEpCAM^loT1alpha^loCD45^hi). Endothelium and leukocytes showed no differences between uninjured and injured groups with regards to fluorescent intensity. ** P<0.01.

TNF p55 receptor signaling triggers early alveolar epithelial dysfunction during lung injury

The observed rapid development of pulmonary edema with increased barrier permeability is the prominent feature of ALI/ARDS, implying damage to the alveolar epithelium and/or pulmonary endothelium. The results of the caspase-8 localization analysis confirm that the impact of TNF p55 death signaling is focused on the type 1 AECs. As epithelial injury is a critical determinant of ALI pathophysiology, we specifically evaluated the involvement of the alveolar epithelium by also measuring AFC to assess its physiological function, and measuring BALF RAGE, a biomarker for type 1 AEC injury (27).

In WT mice, acid aspiration produced a significant (49%) deterioration in AFC (figure 5A) and upregulation in BALF RAGE (Figure 5B), as compared to uninjured animals. There were strong inverse correlations between AFC and BALF RAGE levels (figure 5C; Pearson r = −0.901; P<0.0001) and between AFC and respiratory system elastance (figure 5D; Pearson r = −0.908; P<0.0001) in WT animals, confirming that the AFC measurements indeed reflect the functional status of the injured alveolar epithelium. Importantly, the decreases in AFC and increases in RAGE were considerably attenuated in p55^−/− mice, suggesting that p55 signaling exerts substantive impact on alveolar epithelial dysfunction in acid-induced ALI.

Figure 5.

Alveolar fluid clearance (AFC) is a good indicator of epithelial function and is maintained in p55^−/− mice. A) AFC is significantly disrupted at an early stage after acid instillation (at 90 minutes), but p55^−/− mice show significantly better AFC than WT animals suggesting improved epithelial function (N=6-9 per group). B) The receptor for advanced glycation end-product (RAGE) is released early in response to injury and p55^−/− mice show reduced alveolar RAGE at 90 minutes (N=5-8 per group). C) Ex-vivo measurement of AFC shows a strong inverse correlation to alveolar soluble RAGE levels after acid aspiration (N=14; Pearson r = −0.901; P<0.0001). D) AFC shows a strong inverse correlation to measurements of respiratory system elastance at 90 minutes (N=14, Pearson r = −0.908; P<0.0001). ***P<0.001; *P<0.05.

Caspase-8 activation dictates the functional ability of the alveolar epithelium to clear fluid

In the absence of actual cell death, we sought to clarify the causal link between activation of death signaling per se and alveolar epithelial dysfunction, both induced by TNF p55 receptor ligation. Hence, we administered a caspase-8 specific inhibitor Z-VAD-IETD (4mg/kg in DMSO, i.v.) just before acid instillation, and measured AFC at 90 minutes in WT animals. We found that there was a marked improvement in AFC (figure 6A) and oxygenation (figure 6B) in mice treated with the inhibitor, compared to vehicle (DMSO)-treated controls. Furthermore, lung caspase-8 activity showed a strong inverse correlation with AFC (Pearson r = −0.843; P<0.0001) (figure 6C), implying that caspase-8 activation has a fundamental influence upon epithelial function in early ALI.

Figure 6.

Caspase-8 activation determines epithelial function during the early stages of lung injury. A) The intravenous administration of a caspase-8 specific inhibitor (Z-VAD-IETD) rescues deteriorations in AFC at 90 minutes induced by acid instillation. B) Caspase-8 inhibition also improves arterial oxygenation (N=5-6 per group, P<0.05 by t-test of end-point 90 minute values). C) The extent to which caspase-8 is activated (or inhibited) shows a strong inverse correlation with AFC (N=16; Pearson r = −0.843; P<0.0001) suggesting that caspase-8 activity is a critical determinant of epithelial function in lung injury. *P<0.05.

Alveolar macrophage-derived TNF is required for caspase-8 induced alveolar epithelial dysfunction

Resident alveolar macrophages are prominent early producers of TNF and their depletion attenuates acid-induced ALI (28). To investigate their contribution to the development of alveolar epithelial dysfunction, we depleted their population to ~10% by intratracheally administering clodronate liposomes (48 hours prior to acid instillation) in WT mice (figure 7A). The increase in BALF TNF levels observed at 90 minutes after acid instillation in control animals (pre-treated with PBS liposomes) was, as expected, virtually abolished in alveolar macrophage-depleted mice (pre-treated with clodronate liposomes) (figure 7B).

Figure 7.

Resident alveolar macrophages initiate TNF/p55/caspase-8 signaling and mediate early epithelial dysfunction in lung injury (N=5-6 per group). A) Resident alveolar macrophages were identified in right lung homogenate as CD45^hiF4/80^hiCD11c^hiCD11b^lo population with high auto-fluorescence. The intratracheal instillation of clodronate encapsulated liposomes produced a 90% depletion of the resident alveolar macrophage population. B) Alveolar TNF is upregulated early (by 90 minutes) in the course of acid induced lung injury and is attenuated by macrophage depletion. C) Macrophage depletion significantly attenuates activation of lung caspase-8 signaling after acid instillation. D) Alveolar levels of soluble receptor for advanced glycation end-products (RAGE) are significantly reduced in macrophage depleted animals implying reduced epithelial injury. E) Macrophage depletion improves lung alveolar fluid clearance (AFC) and thereby improvements in oxygenation (F) (P<0.05 by t-test of end-point 90 minute values). A strong inverse relationship is noted between G) AFC and caspase-8 (N=17; Pearson r = −0.845; P<0.0001; uninjured data points are the same as for figure 6C) as well as H) AFC and RAGE (N=17; Pearson r = −0.894; P<0.0001; uninjured data points are the same as for figure 5C), implying that epithelial death signaling and dysfunction determine AFC in acid induced lung injury. ***P<0.001; **P<0.01; *P<0.05.

This depletion of alveolar macrophages markedly attenuated the activation of caspase-8 within the lung (figure 7C), to levels similar to those found in the p55^−/− animals. Interestingly, we found no detectable soluble FasL in BALF of acid injured animals, suggesting that death signaling and caspase-8 activity at this time point (90 minutes) was driven predominantly by TNF. Macrophage depletion also produced a significant reduction in BALF RAGE levels (figure 7D), and substantial (1.8 fold) improvement in AFC (figure 7E), ultimately leading to improved oxygenation (figure 7F). Once again, there was a strong inverse correlation between caspase-8 activity and AFC (Figure 7G; Pearson r = −0.845; P<0.0001) and between RAGE and AFC (Figure 7H; Pearson r = −0.894; P<0.0001).

DISCUSSION

In this study, we demonstrated for the first time a crucial role of TNF-mediated ‘death signaling’ in regulating alveolar epithelial dysfunction and development of pulmonary edema during the early phase of ALI. Using an in vivo mouse model of acid-induced ALI, we found that alveolar macrophage-derived TNF promotes pulmonary edema formation, specifically through activation of the TNF p55 receptor and its downstream death signaling. The resultant caspase-8 activation in type 1 alveolar epithelial cells dictates lung epithelial injury and dysfunction, as assessed by AFC and BALF soluble RAGE levels. Our findings support a novel concept that TNF p55 receptor-mediated death signaling per se produces significant dysfunction in the alveolar epithelium, and this effect, rather than actual apoptotic cell death, determines the pathophysiology and physiological derangement in early ALI.

Using p55 deficient animals, we found that TNF, through its p55 receptor, plays a major role in the early alveolar epithelial injury following acid aspiration. TNF signals through two cell surface receptors, i.e. TNF receptor 1 (p55, TNFRSF1a, CD120a) and TNF receptor 2 (p75, TNFRSF1B, CD120b). The majority of responses to TNF are mediated via p55 signaling, although it has become clear that p75 signaling may have independent, even opposing effects. The current study focuses on the p55 receptor because in cases where such differential signaling has been demonstrated, p55 is universally the ‘deleterious’ receptor (20, 29, 30), and thus much more amenable to development of pharmacotherapies (31). Ligand binding to p55 allows the assembly of TNF receptor complex I on the cell membrane leading to NFκB and MAPK-mediated gene transcription (32). This is at the heart of the classic roles of TNF ‘proinflammatory’ signaling in promoting production of downstream inflammatory cytokines/chemokines and migration of leukocytes (33). However, our results indicate that the injurious effect of p55 signaling on alveolar-capillary barrier dysfunction is not dependent on the classic inflammatory consequences of TNF, consistent with our previous report using a mouse model of ventilator-induced lung injury (20), suggesting that the current findings are not model-specific.

The alternative pathway to this proinflammatory signaling is so-called ‘death signaling’, which involves internalization of complex I leading to formation of TNF receptor complex II and subsequent recruitment of the death domain proteins, TNF Receptor Associated Death Domain and Fas Associated Death Domain (34). Crucially, death signaling is only induced through the p55 TNF receptor, and not the p75 receptor. The mechanisms behind the decision making process between complex I and II remain unclear, although there is evidence that ability of complex I to activate NF-κB may shift the cells towards an anti-apoptotic proinflammatory direction (35). Complex II helps recruit and activate caspase-8, a key modulator of death signaling, which ultimately activates executioner caspases, such as caspase-3, promoting cell death. Here we found early activation of p55 receptor-mediated death signaling involving caspase-8 after acid aspiration, and obtained clear in vivo evidence that this death signaling, rather than classical proinflammatory cascades, plays a crucial role in determining physiological dysfunction of the epithelium, as assessed by impaired AFC and elevated BALF soluble RAGE levels.

It is important to note that alveolar epithelial dysfunction manifested in this study long before significant amounts of apoptotic cell death took place. Even by 3 hours TUNEL positive cell numbers were very limited, indicating that the substantial lung dysfunction observed cannot be attributed simply to epithelial cell loss. Although efficient clearance of dead cells by macrophages, as seen in reports studying in vivo phagocytosis of exogenously applied apoptotic bodies (36), may explain this scarcity, it is not likely for both the end-stage apoptosis and efferocytosis processes to take place at this early time point (90-180 min after injury). Furthermore, there were no histological signs of disruption of alveolar structure or proliferative response that are expected to occur following apoptosis and removal of structural cells such as alveolar epithelium (37). As described before, the extent of morphological apoptosis (i.e. TUNEL positive cells) in ARDS non-survivors is in general limited, i.e. up to 10%, which includes apoptotic neutrophils (13), and the majority of animal ALI models at the peak of injury show minimal levels of TUNEL positive cells (38, 39). This situation is very similar to the one in chronic heart failure, where despite the convincing evidence of apoptosis involvement in the disease progression, the majority of cardiac myocytes do not show nuclear damage (i.e. are not yet dead) and with only changes to cytoplasmic apoptotic signaling, still present significant systolic dysfunction (40). These discrepancies between the extent of ‘terminal’ apoptotic cell death and physiological indices of cell/organ dysfunction have not attracted proper attention in ALI research. The present study provides clear evidence that caspase-8 activation within the type 1 alveolar epithelium determines the degree of impairment in AFC and oxygenation, supporting a new concept that TNF-induced death signaling itself, rather than the number of dead cells, determines alveolar epithelial dysfunction and injury in the early phase of ALI.

A number of mediators have been shown to influence AFC in animal and clinical studies (41, 42). TNF has been proposed to have an opposing impact on AFC with receptor binding inducing an inhibition and its lectin-like domain promoting reabsorption (43). This study is the first to implicate death signaling having a direct influence upon AFC. However, the mechanism by which this occurs is unclear. One possibility is that localization of activated caspase-8 to the mitochondrial membrane (44) may impede mitochondrial function and reduce ATP production necessary for basolateral sodium-potassium ATPase transporter function. Alternatively, TNF may down-regulate apical epithelial Na channel expression or function, as observed in type 2 pneumocytes in vitro (45), through further downstream consequences of death signaling, e.g. phosphorylation of myosin light chain kinase, which has been shown to induce dynamic cell membrane blebbing through increased cytoskeletal contractility (46). Indeed, hypoxia has been shown to induce alveolar epithelial cytoskeletal disruption and reduce epithelial Na channel expression, which are recovered by a pan-caspase inhibitor (47), suggesting that death signaling may impact on ion channel expression/activity. The mechanisms behind this novel role of TNF death signaling in regulating alveolar fluid dynamics remains to be fully explored.

The results of alveolar macrophage depletion experiments, which confirmed the previously reported importance of alveolar macrophages in acid-induced ALI (28), showed decreased TNF production as expected, along with attenuated lung caspase-8 activation, reduced BALF RAGE levels, improved AFC, and improved physiological indices (i.e. oxygenation). Importantly, in acid instilled PBS liposome-treated mice, we could not detect soluble Fas ligand within the alveolar space whereas a substantial increase in soluble TNF was detected. However, there is a possibility that membrane bound FasL may activate Fas on epithelial cells leading to caspase-8 activation. This may explain why p55 receptor knockout animals show only a 60% reduction in caspase-8 activation compared to WT animals. Although we have not completely excluded the involvement of cell surface FasL from alveolar macrophages, Fas-induced lung injury (utilizing the anti-Jo2 antibody) has been shown to worsen, rather than improve following alveolar macrophage depletion (48). Of note, the majority of literature reports alveolar FasL/Fas activation at later time-points (>4-6 hours), potentially as a consequence of Fas ligand release by blood-derived leukocytes (14, 49, 50).

There is growing evidence that apoptosis may not be as simple and straightforward a process as once considered (51): an injured cell may enter a state of ‘apoptotic limbo’ i.e. always on the edge of death, depending on, for instance, a threshold of initial caspase activation (52) or the availability of cellular ATP for apoptosis to follow through to completion (53). Prior to this threshold, cell death may be reversible, but once a critical threshold is reached, nuclear damage is inevitable, ultimately leading to irreversible cell death. In ALI, this may be dependent on the extent to which death signaling is initially activated by early phase ligands such as TNF from alveolar macrophages, and subsequently perpetuated by later phase mediators such as Fas driven by infiltrating leukocytes. These considerations suggest new early-phase therapeutic strategies for ALI/ARDS – targeting early death signaling, e.g. specific blockade of p55 receptor (31) or caspase-8 activation, which may result in recovery of alveolar epithelial dysfunction and injury.

There are some important caveats to our work. Whilst the acid aspiration model is a widely utilized model of ALI and often quoted as most clinically-relevant among the currently available ALI models (54, 55), it has many limitations: for instance, it utilizes acid solutions with much lower pH than that of aspirated gastric contents, and clinical aspiration includes other particulate matter such as food and bacterial products. Thus, direct extrapolation from this model to clinical ARDS should be done with caution. However, the aim of our model was to produce significant physiological deterioration (acquiring oxygenation indices similar to clinical ARDS) enabling us to assess the mechanisms of alveolar epithelial dysfunction. Moreover, we have shown previously that deletion of the p55 receptor leads to similar protective effect, seemingly also independent of downstream inflammation, in a model of stretch induced lung injury (20). Collectively, these findings strongly suggest that TNF-mediated death signaling has wider implications to the pathogenesis of epithelial dysfunction in lung injury of various etiologies.

In conclusion, we present a critical role for early intrapulmonary death signaling and alveolar epithelial dysfunction in ALI, mediated by alveolar macrophage-derived TNF producing caspase-8 activation through its p55 receptor. This activation of death signaling within alveolar epithelial cells, rather than cellular loss through completed apoptosis, critically regulates the ability of the alveolar epithelium to resolve pulmonary edema through alveolar fluid clearance. The work provides a coherent link to a number of unresolved paradigms implicated in the early pathogenesis of ALI: integrating the injurious role of alveolar macrophages; TNF; epithelial injury and apoptosis; and alveolar fluid clearance. Finally, the results provide a basis to consider translational potential for novel strategies targeting alveolar epithelial death signaling and dysfunction to treat ALI/ARDS.

Abbreviations

AEC

alveolar epithelial cell

AFC

alveolar fluid clearance

ALI

acute lung injury

ARDS

acute respiratory distress syndrome

BALF

bronchoalveolar lavage fluid

Fas

fas receptor

FasL

fas ligand

RAGE

receptor for advanced glycation end-products