Hypoxemic respiratory failure of the neonatal organism involves increased acid sphingomyelinase (aSMase) activity and production of ceramide, a second messenger of a pro-inflammatory pathway that promotes increased vascular permeability, surfactant alterations and alveolar epithelial apoptosis. We comparatively assessed the benefits of topical aSMase inhibition by either imipramine (Imi) or phosphatidylinositol-3,5-bisphosphate (PIP2) when administered into the airways together with surfactant (S) for fortification. In this translational study, a triple-hit acute lung injury model was used that entails repeated airway lavage, injurious ventilation and tracheal lipopolysaccharide instillation in newborn piglets subject to mechanical ventilation for 72 hrs. After randomization, we administered an air bolus (control), S, S+Imi, or S+PIP2. Only in the latter two groups we observed significantly improved oxygenation and ventilation, dynamic compliance and pulmonary oedema. S+Imi caused systemic aSMase suppression and ceramide reduction, whereas the S+PIP2 effect remained compartmentalized in the airways because of the molecule's bulky structure. The surfactant surface tensions improved by S+Imi and S+PIP2 interventions, but only to a minor extent by S alone. S+PIP2 inhibited the migration of monocyte-derived macrophages and granulocytes into airways by the reduction of CD14/CD18 expression on cell membranes and the expression of epidermal growth factors (amphiregulin and TGF-β1) and interleukin-6 as pro-fibrotic factors. Finally we observed reduced alveolar epithelial apoptosis, which was most apparent in S+PIP2 lungs. Exogenous surfactant “fortified” by PIP2, a naturally occurring surfactant component, improves lung function by topical suppression of aSMase, providing a potential treatment concept for neonates with hypoxemic respiratory failure.
Hypoxemic respiratory failure of the neonate is a life-threatening manifestation of neonatal acute lung injury (nALI) that is defined as severely impaired lung function involving reduced lung compliance, permeability oedema, increased surfactant surface tension, transpulmonary migration of macrophages and alveolar epithelial apoptosis. A variety of clinical conditions ultimately cause nALI including (but not limited to) congenital pneumonia, sepsis, meconium aspiration and pulmonary haemorrhage, all of which are characterized by severe inflammation. An effective approach to pharmacological treatment remains to be elucidated. Severe hypoxemia may develop despite mechanical ventilation and surfactant replacement therapy, which results in an unacceptably high rate of mortality among the babies affected.
In nALI models, the physical properties of surfactant films are impaired by increased concentrations of the phospholipid ceramide [1, 2]. Ceramide is generated by the enzyme acid sphingomyelinase (aSMase), which catalyses the degradation of cellular sphingomyelin to phosphorylcholine and ceramide. Ceramide inhibits the synthesis of dipalmitoyl- phosphatidylcholine (DPPC) by inhibiting the key enzyme cholinephosphate cytidylyltransferase . The addition of ceramide to surfactant increases the surface tension of surfactant films in a dose-dependent manner . Ceramide augments permeability oedema which further contributes to surfactant dysfunction . Finally, ceramide mediates caspase-dependent  apoptosis, as evidenced in lung fibroblasts and airway epithelial cells .
L-α-phosphatidyl-D-myo-inositol-3,5-bisphosphate (PIP2) is a phosphoinositide that is important for stress signalling in organelles. PIP2 is rapidly derived from the ubiquitous surfactant component phosphatidylinositol by sequential phosphorylation by phosphoinositide kinases . PIP2 is synthesized and excreted by type II pneumocytes in the lung , but it represents only a small portion of the total cell phosphoinositides (<0.1%). Most notably, PIP2 is the most potent aSMase inhibitor known to date, as demonstrated in cell culture studies [10, 11].
In a previous study we demonstrated the benefits of the non-specific aSMase inhibitor imipramine (Imi) for gas exchange, lung mechanics and pulmonary oedema in a nALI model of repeated airway lavage . Because Imi has a high volume of distribution, its use in neonates may be perilous as a result of the systemic suppression of the critical aSMase-ceramide signal transduction pathway. Therefore, we evaluated topical PIP2 admixed with surfactant as an alternative because (i) it is produced naturally in the body, and (ii) it has a bulky molecular structure, which precludes distribution beyond the airway compartment.
In the present study, we used a neonatal porcine triple-hit lung injury model (repeated airway lavage, injurious ventilation and tracheal lipopolysaccharide instillation) throughout 72 hrs of mechanical ventilation. We focused on the comparison of the two test substances, Imi and PIP2 admixed with exogenous surfactant, and assessed the generalized non-specific Imi effect compared to the topical alveolar epithelial aSMase inhibition mediated by PIP2.
Materials and methods
The experimental protocol was approved by the review board of the Schleswig-Holstein government for the care of animal subjects (letter of acceptance V312-72241.121-24) in accordance with the German law for animal protection (BGBI 1, page 1319), and the European Community guidelines (2007/526/EC). A total of 30 piglets were studied between days 2 and 6 of life.
After sedation and intubation, mechanical ventilation was provided by pressure-limited neonatal ventilators with the following settings: FiO2 = 0.5, PEEP = 6 mbar, and rate = 25/min. The peak (inspiratory pressure) was adjusted every hour to maintain the tidal volume (VT) at 7 ml/kg. PaCO2 was maintained within a range of 35–50 mmHg by controlling the rate and PaO2 within 50–150 by controlling the FiO2.
Repeated broncho-alveolar lavage (rBAL, first step of nALI) with 30 ml/kg saline was continued until the PaO2/FiO2 decreased to ~100 mmHg (baseline). At 25 hrs, the piglets were submitted to a 2 hrs-period of injurious ventilation (second step) with zero-PEEP ventilation followed by doubled VT (14 ml/kg). In a third step, at 49 hrs, 2.5 mg lipopolysaccharide (E. coli serotype O127:B8) in 0.5 ml saline was instilled into the trachea.
The piglets were randomized to one of the following four groups:
control (C) piglets received an air bolus;
surfactant (S) piglets received 50 mg/kg surfactant (poractant alfa, Chiesi Farmaceutici, Parma, Italy);
S+Imi piglets received 5 mg Imi admixed with surfactant;
S+PIP2 piglets received 2 mg PIP2 (#10008398, Cayman, Tallinn, Estonia) admixed with surfactant.
Following each lung injury protocol the group-specific intervention was carried out three times (at 2, 26 and 50 hrs).
We calculated the oxygenation index [OI: MAP (mean airway pressure) × %O2/PaO2] and a ventilation efficiency index [VEI: 3800/(Peak-PEEP) × f × PaCO2]. Calculations of the extra-vascular lung water index (EVLWI, ml/kg) and cardiac index (CI, l/min./m2) were performed using the transpulmonary indicator dilution technique . The single-breath least square method was used to assess the dynamic (specific) compliance (sCrs) and resistance (Rrs) of the respiratory system as previously described .
Prior to killing the piglets a final diagnostic BAL (dBAL) was carried out; lung tissue specimens were then gathered for the subsequent analyses.
BALF cell counts and differentials: BALF was filtered to remove gross particles and secretions from the airways, and was centrifuged for 4 min. at 4°C and 20 × g to separate cells. Cellular material was resuspended into phosphate-buffered saline for cell counts, cell differentiation by standard Romanowsky staining, and for flow cytometry. For flow cytometry the cell pellet was resuspended in 30 μl blocking reagent and 120 μl washing buffer. After a washing step, surface staining with a combination of FITC- (CD14 antigen, IgG2a) and PE- (CD18 antigen, IgG1) conjugated monoclonal antibodies and their appropriate isotype controls (all reagents from antibodies-online, Aachen, Germany) was achieved by incubation of ~105 cells with the monoclonal antibody cocktails for 30 min. at 4°C in the dark. Cells were subsequently analysed using flow cytometry using a FACSCalibur and the CellQuest software (Becton Dickinson, Heidelberg, Germany).
Surfactant was isolated by filtering and centrifuging BALF for 5 min. at 4°C and 20 × g. The pellet was then weighed and resuspended in normal saline at a concentration of 10 mg/ml thus assessing surfactant quality only and disregarding the individual surfactant pool in the piglets. Surfactant in aliquots of 0.5 ml was put on the surface of a warmed normal saline trough for determination of minimal (and maximal) surface tension by the use of a modified Wilhelmy balance (E. Biegler, Mauerbach, Austria).
Real-time qPCR for the quantification of amphiregulin, TGF-β1, IL-6:RNA isolation was performed with 30 mg lung powder with NucleoSpin RNA II Kit (Machery Nagel, Düren, Germany) automated on QIAcube roboter (Qiagen, Hilden, Germany). RNA quantification and quality control was done using a NanoDrop 1000 spectrophotometer (Thermo Fisher, Waltham, MA, USA). 675 ng of RNA, 1 μl oligo(dT)15 primer (0.5 μg/μl; Invitrogen, Karlsruhe, Germany) were added in a total volume of 12 μl of H2O. After 10 min. incubation at 65°C on a UNO II thermocycler (Biometra, Göttingen, Germany), five times 4 μl buffer, 2 μl dNTP (10 mM), 1 μl RNasin (40 U/μl), and 1 μl M-MLV RT  (200 U/μl; all reagents from Promega, Mannheim, Germany) were added. This was followed by two incubation periods of 90 min. at 40°C, and 5 min. at 95°C, respectively, on UNO II thermocycler. Real-time qPCR: 1 μl cDNA as template was incubated with 312.5 nM forward primer (248 nM for Il-6; Eurofins, Ebersberg, Germany), 312.5 nM reverse primer (248 nM for Il-6) and SYBR-Green I Mastermix (Roche Diagnostics, Mannheim, Germany) in a LightCycler480 (Roche). Cp values were acquired by the Second Derivative Maximum method. Advanced relative quantification was performed using the LightCycler 480 Software 1.5 SP3 (Roche), and efficiency corrected by in-run standard curves using the Roche Applied Science E-Method . Data were referenced to the correspondent housekeeping gene ß2-microglobulin and normalized to the mean of the C-group values. For quality control, in-run negative controls, melting curve profiles and product separation in agarose gels were performed.
aSMase activity and ceramide concentrations in pulmonary tissues, BALF and serum: aSMase activity were assayed in a modified micellar in vitro assay using sphingomyelin labelled with carbon 14 (14C) (Amersham Pharmacia Biotech, Piscataway, NJ, USA) as substrate, as described by Wiegmann et al. . Frozen lung tissue samples were pulverized and melted, of which 10 μg of protein was used to assay aSMase activity in the presence of 1.4 mM ZnCl2 in a buffer (final volume of 50 μl) containing 250 mM sodium acetate, 1 mM EDTA (pH 5.0) and 2.25 μl of [N-methyl-14C]sphingomyelin. Zn2+ dependency was assayed by replacing Zn2+ by 1 mM EDTA. Phosphorylcholine was then extracted with 800 μl of chloroform/methanol (2:1 [vol/vol]) and 250 μl of H2O and the amount of radioactive phosphorylcholine produced from hydrolysis of 14C-sphingomyelin determined in the aqueous phase by scintillation counting. For aSMase activity measurements in fluids (BALF and serum), initially 30 μl fluid were mixed with 280 μl aSMase-buffer, and identical steps were taken for final scintillation counting.
Ceramide levels were determined as described by Jensen et al.  after extraction from homogenized tissue by chloroform/methanol, separation of lipids by high performance thin-layer chromatography, and quantification by two-dimensional charring densitometry.
TUNEL staining for the quantification of alveolar epithelial apoptosis: In situ-tailing was applied to visualize DNA fragmentation in apoptotic cells (#11684817910, Roche Applied Science). Sections were deparaffinized, rehydrated, boiled in citrate buffer and immersed in TritonX-100/hydrogen peroxidase. Non-specific staining was reduced by incubating sections with 3% bovine serum albumin/Tris buffered saline for 1 hr in room temperature before applying the TUNEL reaction mixture (1 hr, 37°C in humidified chambers). After washing, converter-peroxidase was incubated on the slides for 30 min. at 37°C. Labelling was visualized by adding DAB substrate (Roche) for 10 min. Negative controls were performed by omitting the TUNEL reaction mixture during the labelling step, which resulted in absence of specific staining. Apoptosis was assessed by light microscopy using an alveolar epithelial apoptosis score: 4–6 alveoli per section were chosen to determine the number of apoptotic cells in an undisturbed alveolar epithelial lining of 200 cells as a reference. The colleague performing the score (SP) was blinded to the sections' group affiliation.
Immunohistochemical detection of (active) caspase-8 : Paraffin-embedded lung specimens were sectioned at 1 μm. Before staining sections were deparaffinized, rehydrated, washed in Tris buffered saline and boiled in fresh citrate buffer (pH 6) using microwave for 10 min. Endogenous peroxidase activity was blocked, and sections were permeabilized by incubation in 0.3% TritonX-100/1% hydrogen peroxidase for 20 min. Sections were then blocked with 10% donkey serum (Dinova, Hamburg, Germany) for 1 hr at room temperature, and the antibody against (active) caspase-8 (rabbit-IgG; Cell Signalling, New England Biolabs, Frankfurt, Germany) was incubated at 4°C overnight. Signals were amplified using the avidin-biotin-complex method (Vectastain peroxidase kit, Vector Laboratories, Burlingame, CA, USA) and visualized with diaminobenzidine. Finally, sections were counterstained with Hämalaun (Sigma, München, Germany), dehydrated in an ascending row of ethanols before mounting with Eukitt (Kindler, Freiburg, Germany), and cover-slipped for assessment using light microscopy.
Non-parametric data (cell differentials, growth factors, cytokines and apoptosis score) were checked for heteroscedasticity and, when necessary, corrected using the Box-Cox transformation before analysis. The differences between groups were analysed by an univariate anova followed by Dunnett's post-test only when the omnibus test showed significant treatment effects. For single-group comparisons, we used two-tailed unpaired t-tests. A repeated measures two-factorial anova for the clinical parameters (Figs 2 and 3), with time and treatment (“treat”) as independent variables, was used to highlight the intermediate-term effects of the specific interventions at the following times: 12–24, 36–48 and 60–72 hrs.
All data represent the means ± SEM unless otherwise specified. P <0.05 was considered to be significant. All analyses were performed using GraphPad Prism 5 and JMP 9.0.
An outline of the experimental protocol is provided in Figure 1.
Two piglets in the S+Imi group did not complete the entire study period and were excluded from the data analysis. Both were diagnosed with acute renal failure (creatinine 152 μmol/l, potassium >10 mmol/l at 52 hrs; oliguria and potassium >9 mmol/l at 42 hrs).
To establish comparability of the four groups at the start of the study (“intact lung”), we confirmed that age, weight, initial PaO2/FiO2 and Peak (inspiratory pressure), as well as the number of lavages needed to achieve a PaO2/FiO2 ~100 mmHg and a Peak ≥19 m bar, and the loss of lavage fluid in the airways were similar (Table 1).
Table 1. Group comparisons before repeated airway lavage (“intact lung”)
C (n = 6)
S (n = 6)
S+Imi (n = 8)
S+PIP2 (n = 8)
Peak = peak inspiratory pressure. Data represent the means ± SD. Univariate anova.
4.3 ± 0.7
3.8 ± 1.5
4.2 ± 0.4
3.7 ± 0.8
2.5 ± 0.4
2.8 ± 0.2
2.6 ± 0.1
2.4 ± 0.4
331 ± 88
422 ± 72
397 ± 75
359 ± 75
13.2 ± 1.9
12.6 ± 2.0
13.7 ± 2.1
11.8 ± 1.9
13.5 ± 2.7
16.0 ± 3.2
16.6 ± 3.4
17.0 ± 4.6
Loss of lavage fluid (%)
10.2 ± 2.6
9.6 ± 3.3
9.2 ± 2.5
9.5 ± 2.9
The clinical parameters OI, VEI, sCrs and EVLWI (Figs 2 and 3) were compared by selecting three time intervals (12–24, 36–48 and 60–72 hrs), which were chosen to assess the intermediate-term (not short-term effects) effects of the interventions. A summary of treatment effects is provided in Table 2 and proves the superiority of S+PIP2 treatment for all four parameters when compared to S alone. EVLWI effects are not apparent before 60–72 hrs. sCrs and EVLWI are equally influenced by S+Imi and S+PIP2. To assess alveolar stability, the “25 hrs” time point after completion of the 1 hr interval of zero-PEEP ventilation (first part of the second protocol of nALI, injurious ventilation) was compared among groups, and the “72 hrs” time point at study end was evaluated similarly, reflecting all lung injury protocols and all interventions to some extent. Significant differences between the groups were found for OI (25 hrs P < 0.01, 72 hrs P < 0.001; univariate anova), VEI (25 hrs P < 0.05), sCrs (72 hrs P < 0.001) and EVLWI (72 hrs P < 0.05).
Table 2. Summary of treatment effects on clinical parameters
S versus S+PIP2
S+Imi versus S+PIP2
P-values of treatment effects (“treat” in Figs 2 and 3) for the four clinical parameters comparing all four groups by repeated measures two-factorial anova. “Overall” indicates the combination of the three intervals 12–24, 36–48 and 60–72 hrs. Single group comparisons (S versus S+PIP2 and S+Imi versus S+PIP2) of all three intervals. The second factor “time” as well as interactions of “treat” and “time” were not significant for either parameter.
Surfactant surface tensions
Minimum surface tensions of surfactant from BALF varied significantly (Fig. 4). To demonstrate the ability of our system to lower surfactant surface tension (a Wilhelmy balance) we confirmed that the addition of 5 mg poractant alfa onto the surface of the saline bath yielded a minimum tension of 19.0 ± 0.1 mN/m, and the addition of 10 mg a tension of 5.9 ± 0.8 mN/m.
Cell concentrations, differentials and cell subsets in BALF
The total cell count in BALF varied significantly between groups after 72 hrs of mechanical ventilation (Fig. 5A). The average pre-lavage cell concentration was approximated by the intervention with S+PIP2 only (130 ± 42 × 103 cells/μl versus 149 ± 46 × 103). The BALF cell differentials showed decreased percentages of granulocytes and increased percentages of monocytes after aSMase inhibition, an effect that was only significant for S+PIP2 treatment (Fig. 5B).
Lung injury induced a two- to fivefold increase in CD14+ and/or CD18+ cells (Fig. 5C, C group). Notably, S+Imi and S+PIP2 significantly reduced the number of CD14+/18+ cells, whereas CD14−/CD18+ cells were only reduced by S+PIP2 intervention (P < 0.01).
Amphiregulin, TGF-β1, and IL-6
Surfactant alone (S) produced equal or slightly increased gene expression compared to that observed in the C group, and only S+PIP2 significantly decreased gene expression to a remarkably low level (Fig. 6).
aSMase activity and ceramide
Compared to the C group, S+Imi reduced aSMase activity in pulmonary tissues by 41.9% (Fig. 7A). There was a 33.2% reduction in the levels of ceramide (Fig. 7B). Because the reduction in aSMase activity in S+PIP2 treated lungs was similar to the S-treated lungs, and no additional effect on aSMase inhibition could be detected, we also compared aSMase activities in the alveolar (Fig. 7C) and blood (Fig. 7D) compartments. The results show that 1. the local alveolar inhibition was even more pronounced (but not significantly different) after treatment with S+PIP2 compared to S+Imi, and 2. only S+Imi, but not S+PIP2, exerted a systemic effect by balancing the aSMase activity levels in serum over time.
TUNEL staining, apoptosis score, and caspase-8 immunohistochemistry
Assessment of alveolar epithelial apoptosis revealed a significant reduction in S+PIP2 treatment compared to S treatment alone (Fig. 8A). Graphic representatives of cells with alveolar apoptosis by TUNEL staining (Fig. 8B) and immunohistochemical staining of active caspase-8 (Fig. 8C) are provided by means of an enlargement.
Intervention-related side effects
To investigate any relevant side effects, we also monitored the resistance of the respiratory system, cardiac index and systemic vascular resistance index, fluid and electrolyte balances, hepatic and renal serum parameters and differential blood cell counts over 72 hrs of mechanical ventilation (data not given). Differences between the groups were only observed for urine production (C: 5.8 ± 0.6 ml/kg/h, S 4.4 ± 0.8, S+Imi 4.6 ± 0.4, S+PIP2 3.8 ± 0.4; P = 0.07, univariate anova) and weight gain (C: +0.34 ± 0.02 kg/kg/72 hrs, S + 0.32 ± 0.03, S+Imi + 0.36 ± 0.03, S+PIP2 + 0.47 ± 0.03; P < 0.05).
The four clinical parameters oxygenation (OI) and ventilation (VEI), compliance (sCrs) and pulmonary oedema (EVLWI) (Figs 2 and 3) improved significantly after topical airway epithelial aSMase inhibition when compared to the C and the S groups. Because the beneficial surfactant effects subside more than 6–12 hrs after replacement therapy in inflammatory lung disease (as also demonstrated in the present study), the benefits of clinical application in term neonates , in children  and especially in adult patients  suffering from ALI/ARDS are not convincing. Additional treatment strategies will be necessary to restore epithelial and endothelial functional integrity.
PIP2: aSMase inhibition and metabolism
Lipid rafts on the membranes of pulmonary epithelial and endothelial cells are constituted of cholesterol and sphingolipids (sphingomyelin and glycolipids) and are located in the exoplasmic leaflet of the plasma membrane . These rafts have a diameter of ~50 nm, corresponding to ~3500 sphingomyelin molecules and a limited set of structural proteins (~10–30) [21, 22]. Thus, >75% of cellular sphingomyelin is concentrated in the rafts . They also contain the enzyme aSMase (effector, Fig. 9), which is subject to rapid activation following stress and produces ceramide from sphingomyelin. Developmental changes increase the levels of aSMase activity and ceramide content in the foetal and neonatal lung when compared to the adult lung as evidenced in rats .
Ceramide and its downstream products sphingosine and sphingosine-1-phosphate are recognized as second messengers and ultimately regulate cell growth, viability, differentiation and senescence . Ceramide, a phospholipid, has been proven to increase surfactant surface tension by integration into the surfactant film , to induce permeability oedema by epithelial/endothelial disruption , and to cause apoptosis . As a consequence, pharmacological inhibition of increased aSMase activity by non-specific aSMase-inhibitors such as Imi improved pulmonary functions in selectively perfused murine lungs  and in a neonatal piglet model of respiratory failure . Because Imi exerts a broad range of undesired effects mediated by the induction of lysosomal proteases and has a high volume of distribution (14.5 l/kg), we comparatively assessed the effects of PIP2, a naturally occurring specific aSMase inhibitor without cell membrane penetrating properties.
PIP2 has been shown to suppress aSMase activity by 47% in cell cultures and is regarded as the most potent (IC50 = 0.53 μmol = 0.51 mg/ml) selective aSMase inhibitor known to date . Recent research revealed that PIP2 is less suitable for cell culture research because of its fivefold negative charge and its two long fatty acid chains (16:0/16:0), which cause it to stack in cellular membranes . For clinical use in sick neonates, however, limited tissue penetration and organ translocation of PIP2 are rather desirable for protecting the delicate neonatal organism [25, 26] from systemic immune suppression. To demonstrate a local (compartmentalized) PIP2 effect within the airspaces, we detected lower aSMase activity in the BALF in the S+PIP2 condition compared to S+Imi-treated piglets (Fig. 7C). As extracellular aSMase activity is present in all body fluids except cerebrospinal fluid , the systemic aSMase inhibiting effect found in S+Imi piglets confirms an unrestricted Imi distribution (Fig. 7D). However, given the extracellular location of aSMase in the outer leaflet of the plasma membrane, an inhibitor does not need to be cell permeable .
Phosphorylation of the phosphatidyl-inositol (PtdIns) head group generates eight varieties of phosphoinositides including PtdIns-3-P, PtdIns-4,5-P2 and PtdIns-3,5-P2 (PIP2). PtdIns-3-P serves as a precursor for PIP2, the latter only reaching concentrations that are ~500-fold lower . PtdIns increases by fourfold in the BALF of adult ARDS patients as a result of decreased levels of PtdIns-specific extracellular phospholipase C activity . Whether or not this increase in PtdIns also serves to counteract pulmonary inflammation and increase the percentage in 16:0/16:0-PtdIns (usually only ~3% of PtdIns in human surfactant), the basic compound for PIP2 synthesis, is not known to date. Saccharomyces cells exposed to hyperosmotic medium increase their PIP2-levels 20-fold within 5 min.  and return to normal 20 min. later. The sustained effects of the S+PIP2 emulsion in the present study suggest, however, decelerated PIP2 release and metabolism from the surfactant emulsion when compared to the aqueous solutions used for cell culture research.
Alveolar epithelial apoptosis
Hyaline membranes, protein-rich alveolar exudates, fibroproliferative changes and epithelial apoptosis all contribute to the impaired gas exchange and lung functions in nALI (Fig. 9). Relative to apoptosis, Bern et al.  demonstrated 1–20% caspase-3 positive pre-apoptotic epithelial cells in pulmonary specimens from children who died of hypoxemic respiratory failure. Aside from developmental processes, lung epithelial apoptosis in newborns and infants occurs in conditions such as respiratory tract infection (including LPS-induced nALI), meconium aspiration, exposure to high oxygen concentrations and mechanical stress . Overt apoptosis is observed only in a minority of alveolar epithelial cells in nALI, but gas exchange is impaired in many more cells as a result of dysfunction of membrane pumps, dissolution of tight junctions and reduced ATP levels . Epithelial apoptosis is mediated by mechanical stretch imposed on type II pneumocytes , LPS , and cross-talk between ceramide and the caspases . According to the “neutrophilic hypothesis” , epithelial apoptosis depends on granulocyte apoptosis, clearance of apoptotic granulocytes in airspaces and their phagocytosis by macrophages leading to phenotype change and decreased production of pro-inflammatory mediators. We suggest, however, that the low aSMase concentrations found in the BALF of S+PIP2-treated piglets (Fig. 7C) essentially inhibited migration of granulocytes and monocyte-derived macrophages into the airspaces , suppressed their metabolic activity  and modulated the epithelial inflammatory response [36, 40] (“epithelial hypothesis”).
Transpulmonary cell migration
Migration of the CD14+/CD18+ cell subset (essentially monocyte-derived macrophages and granulocytes) was inhibited by S+Imi and by S+PIP2 treatment. Maus et al. showed in a murine model of MCP1-induced ALI that blood monocytes and resident alveolar macrophages, in contrast to monocyte-derived macrophages, do not express CD14 . Because CD14/LPS-interactions result in cellular signalling and production of pro-inflammatory cytokines, reduced CD14 expression may be an important effector modulation to reduce uncontrolled pulmonary inflammation .
In contrast, CD18 is expressed on all three monocyte-derived cell lines as well as on granulocytes. In LPS-induced lung injury in rabbits, blockade of the adhesion receptor CD18 (β2-integrin) with a monoclonal antibody resulted in reduced granulocyte invasion and permeability edema but did not influence macrophage recruitment . S+PIP2 treatment exerted significant reductions in CD14 and CD18 expressions: the former is an effect of Imi that has been previously described in cell cultures stimulated by LPS or ceramide , the latter represents a new finding, which contributed to the reduced percentages of granulocytes and the reduced cell concentrations in BALF.
Growth factors and IL-6
Amphiregulin is a polypeptide epithelial growth factor receptor (EGFR) ligand in airway epithelial cells, and its expression follows exposure to stressors such as stretch-mediated overventilation [45, 46]. This effect was significantly suppressed by S+PIP2 intervention in our model (Fig. 6A), which also involved a 60-min period of overventilation (at 26 hrs). TGF-β1 (Fig. 6B), a pleiotropic polypeptide involved in both the positive and negative regulation of cell proliferation and differentiation , is recognized as a driving force in fibroproliferation in the acute/subacute phases of ALI/ARDS [48, 49], and its expression is mutually dependent on ceramide generation . The marked reductions in TGF-β1 and amphiregulin mediated by S+PIP2 are novel effects that might improve the prognosis of nALI in human neonates as a result of the protection against fibroproliferative lung repair. Whether suppression of TGF-β1 expression occurred because of the aSMase suppression or by direct PIP2-interactions cannot be deduced from the data of this study. Pekary & Hershman , however, suggested different second messenger signalling pathways for TNF-α-induced aSMase activation and TGF-β1 activation by MAP kinases and the MAD protein pathways, an observation rather favouring a direct PIP2-effect. The marked reduction in IL-6 as a cytokine with universal impact in pulmonary inflammation, atelectasis/alveolar recruitment  and cyclic stretching of endothelial cells , observed in S+PIP2-treated piglets could also contribute to the reduced stimulation of alveolar epithelial cells and macrophages producing collagen fibres .
Surfactant function and pulmonary oedema
Surfactant function (Fig. 4) and pulmonary oedema (Fig. 3B) differed significantly between groups. Although the effect of ceramide on surfactant function is twofold, direct as a result of the integration into the surfactant film , and indirect, e.g. because of the decreased surfactant protein B gene expression , the mechanisms behind oedema generation are less obvious. In all likelihood alterations of the endothelial cells with increased permeability are followed by activation of ceramide-binding proteins and can be inhibited by several serine/threonine protein kinase inhibitors .
Triple-hit neonatal acute lung injury (nALI)
Presuming that the aSMase/ceramide pathway is a central pro-inflammatory pathway in nALI models [5, 56], our intention was to maximize the clinical relevance of our translational model through the application of different local stressors to the pulmonary epithelium/endothelium, each of which is able to induce respiratory failure by itself in piglet nALI models: repeated airway lavage, injurious ventilation and tracheal LPS instillation (Fig. 9, stressors). A more detailed discussion of this model is provided in a previous publication .
Limitations of this study
PIP2 dosage: We used 2 mg PIP2 dissolved in 2.5 ml/kg (~0.32 mg/ml) poractant alfa given into an estimated functional residual capacity of ~20 ml/kg after repeated airway lavage. Mindful of bronchoalveolar lining fluid and extra-alveolar pulmonary oedema, the final concentration of PIP2 may be lower. The only PIP2 dosage for cell culture studies was provided by Kölzer et al.  using concentrations of 0.51 mg/ml to achieve aSMase inhibition of 47%, which is comparable to the reduction in the present study. In a neonatal piglet lavage model , we showed that S+Imi treatment reduced aSMase activity and ceramide content almost to the levels observed in the uninjured lung. Furthermore, aSMase activity reduction is most likely counterproductive and bears the risk of alveolar lipoproteinosis symptoms observed in experimental models of Niemann-Pick disease .
PIP2 compartmentalization: Compartmentalization within the airways was only proven by indirect methods assessing aSMase activity levels in BALF and in serum (Fig. 7C and D). As aSMase is present in all body fluids and tissues , the higher levels of activity in pulmonary tissue (Fig. 7A) and serum suggest limited epithelial penetration of PIP2 when compared to Imi.
aSMase activity: Whether reduced aSMase activity occurred as a result of suppression of activity or because of reduced gene expression by PIP2 was not assessed in our study. To the best of our knowledge aSMase gene expression studies with focus on kinetics have not been performed in clinical or experimental ALI yet. Zhang and Duan, however, showed in a cell culture study using Int407 cells subject to boswellic acid exposure that aSMase activity decreased continuously over 48 hrs, and that reduced aSMase gene expression was detectable as early as 24 hrs of incubation .
Decreased urine production: A S+PIP2 induced reduction in urine output was documented in our study. In a study with isolated basolateral proximal tubular cells exposed to angiotensin, activation of membrane G protein PtdIns-specific extracellular phospholipase Cβ resulted in increased Na+-ATPase activity and hydrolysis of PtdIns-4,5-P2 to diacylglycerol and inositoltrisphosphate; Na+-excretion (and urine production) can be blocked by inhibitors of this phospholipase Cβ or by PtdIns-4,5-P2 .
In the present neonatal triple-hit acute lung injury study we demonstrate that surfactant replacement therapy alone has no sustained effects on gas exchange, lung mechanics and pulmonary oedema. In contrast, an admixture of aSMase-inhibitors “fortifies” surfactant regardless of the more ubiquitous S+Imi effect or the compartmentalized S+PIP2 effect. Notably improved surfactant surface tensions and the lowest degree of alveolar epithelial apoptosis were observed in the S+PIP2 group, which demonstrates that cellular permeability is not necessary for the suppression of alveolar epithelial aSMase activity. As presented here for the first time, S+PIP2 treatment also suppresses gene expression of the epithelial growth factors amphiregulin and TGF-β1, and of IL-6, thus protecting the lung from pro-fibrotic stimuli during lung repair, as well as from the transpulmonary migration of monocyte-derived macrophages and granulocyates through the suppression of CD14/CD18 expression. Surfactant “fortification” by PIP2, a naturally occurring surfactant constituent, modulates inflammation at the alveolar epithelial level, is equal to Imi in terms of aSMase suppression, and exerts important additional anti-inflammatory effects. PIP2 application seems to be an interesting biological treatment concept for neonates with hypoxemic respiratory failure.
The authors thank Jürgen Hedderich, Ph.D., Institute of Medical Informatics and Statistics, for statistical advice. The surfactant preparation was generously provided by Chiesi Farmaceutici, Parma, Italy. This work was supported by a grant from Else Kröner-Fresenius-Stiftung, Bad Homburg, Germany (to M.F.K.), and by intramural funding (to M.F.K.).
S.P., F.D.O., J.S., Sa.S., Pv.B. and M.F.K. executed the clinical experiments. S.P., Pv.B., S.U., St.S. and M.F.K. contributed to the study design. S.P., S.W.-M., S.A.-K., F.K.-L., D.L., D.W. and J.H.-F. designed the post-clinical methods and performed the analyses. M.F.K. conceived the study and wrote the first draft of the manuscript. All authors contributed to the preparation of the manuscript and approved the final version.
Conflict of interests
The authors confirm that there are no conflicts of interest.