Abstract
Alveolar macrophages (AMφ) have been implicated in the polymorphonuclear leukocyte (PMN) recruitment to the lungs during sepsis. Using an in vivo murine model of sepsis (feces in the peritoneum), we show that peritonitis leads to increased activation of AMφ and PMN migration into pulmonary alveoli. To assess cellular mechanisms, an in vitro construct of the pulmonary vascular-interstitial interface (murine AMφ, pulmonary endothelial cells, and PMN) and a chimera approach were used. Using immunologic (Abs) and genetic blockade (CXCR2-deficient AMφ), we show that CXC chemokines in septic plasma are responsible for the activation of AMφ. The activated AMφ can promote PMN transendothelial migration, even against a concentration gradient of septic plasma, by generating platelet-activating factor and H2O2. Platelet-activating factor/H2O2 induce an oxidant stress in the adjacent endothelial cells, an event that appears to be a prerequisite for PMN transendothelial migration, since PMN migration is abrogated across Cu/Zn-superoxide dismutase overexpressing endothelial cells. Using gp91-deficient endothelial cells, we show that NADPH oxidase plays an important role in the AMφ-induced PMN transendothelial migration. Pharmacologic/small interfering RNA blockade of Src kinase inhibits AMφ-induced endothelial NADPH oxidase activation and PMN migration. Collectively, our findings indicate that the PMN transendothelial migration induced by septic AMφ is dependent on the generation of superoxide in endothelial cells via the Src kinase/NADPH oxidase signaling pathway.
One of the most frequent complications of sepsis is acute lung injury (ALI)3/acute respiratory distress syndrome (ARDS) (1, 2). A common pathophysiologic feature of ALI/ARDS is pulmonary inflammation, characterized by an alveolar infiltrate consisting primarily of polymorphonuclear leukocytes (PMN) (1, 2). It is generally thought that the sepsis-induced pulmonary dysfunction/injury is due to the infiltrating PMN (1, 3, 4). PMN emigration from the vascular space to the alveoli is facilitated by the establishment of an interstitial-to-vascular chemotactic gradient (2, 5, 6, 7, 8, 9). The alveolar macrophage (AMφ), with its arsenal of inflammatory mediators (10), is one resident cell admirably suited to fulfill this function. Indeed, AMφ have been proposed as pivotal effector cells in ALI/ARDS responsible for PMN recruitment and vascular protein leakage (2, 11, 12, 13, 14).
Bacterial infections of the lungs and abdomen are the most common causes of sepsis (10, 15). It is readily apparent that bacterial infection of the lungs could directly activate the resident AMφ via release of bacterial products (e.g., LPS) (9, 12, 16, 17, 18, 19). The activated AMφ would release inflammatory mediators, thereby creating an alveolar-to-blood chemotactic gradient to facilitate PMN recruitment into the lung. It is more difficult to envision how AMφ contribute to PMN infiltration of the lungs during sepsis induced by bacterial infection of the abdomen (e.g., peritonitis) (1, 9, 20, 21). In this case, the high levels of circulating inflammatory mediators generated by immune cells (e.g., macrophages, mast cells) in the abdomen (22) would tend to appose the establishment of a chemotactic gradient for PMN recruitment to the lungs. Nonetheless, there are several reports that indicate that sepsis induced by intraperitoneal endotoxin or fecal matter (cecal ligation and perforation; CLP) results in PMN infiltration of the pulmonary interstitium and vascular protein leak (4, 23, 24, 25, 26, 27). Since AMφ have been implicated in the pulmonary vascular leak induced by CLP (14), a major aim of the present study was to assess the mechanisms by which AMφ promote PMN infiltration of the lung during peritonitis.
It is now well recognized that efficient PMN emigration into tissues is dependent on active participation of endothelial cells (28, 29). Endothelial cell redox signaling is thought to be an important initiator of increased surface expression of adhesion molecules that facilitate PMN transendothelial migration (29). We have previously shown that PMN transendothelial migration induced by activated cardiac myocytes is dependent on endothelial generation of superoxide (30). A major source of superoxide that can modulate redox signaling in endothelial cells is NADPH oxidase (29, 31, 32). Furthermore, the LPS-induced transendothelial migration of human monocytic U937 cells is reduced by 35–40% when endothelial cell NADPH oxidase is depleted using a small interfering RNA (siRNA) approach (33). Thus, in the present study we addressed the role of endothelial cell NADPH oxidase in the PMN transendothelial migration induced by sepsis-activated AMφ.
Herein, using an in vitro construct of the pulmonary vascular-interstitial interface (isolated relevant murine cells in cell culture inserts) we provide evidence to support the following two major points. First, peritonitis-induced sepsis results in AMφ activation, such that they promote PMN transendothelial migration against a chemotactic gradient established with septic plasma. Second, the AMφ-induced PMN transendothelial migration is dependent on the generation of superoxide in endothelial cells via the Src kinase/NADPH oxidase signaling pathway.
Materials and Methods
Animals
Cu/Zn superoxide dismutase (SOD) overexpressing mice (Cu/Zn SOD Tg), mice deficient in the gp91 subunit of NADPH oxidase (gp91−/−) on a C57BL/6 background, and BALB/c mice heterozygous for the CXC chemokine receptor 2 (CXCR2−/+) were purchased from The Jackson Laboratory. The BALB/c CXCR2−/+ mice were mated with C57BL/6 wild-type (WT) mice. Subsequently, the new generation of CXCR2−/+ hybrids (BALB/c-C57BL/6) were mated to generate the homozygous CXCR2-deficient (CXCR2−/−) hybrids. This animal study was reviewed and approved by the University of Western Ontario Committee on Animal Care.
Sepsis (peritonitis) was induced by introducing feces into the peritoneum (FIP). In brief, fecal material obtained from the cecum of five to six mice was pooled and suspended in saline (180 mg feces/1 ml saline). Naive mice were injected i.p. with 0.5 ml of the pooled feces. Sham animals were given the same volume of saline. The mice received saline (0.5 ml) for fluid resuscitation 2 h after FIP. For the in vivo studies, 6 or 12 h after FIP, the lungs were removed and indices of inflammation assessed. For the in vitro studies, 6 h after FIP, either plasma (20% dilution) or AMφ were obtained for probing the cellular mechanisms involved in the inflammatory response.
Cells
Pulmonary endothelial cells (PEC).
PEC were obtained from digested lungs of anesthetized adult mice using the microbead technique according to an established protocol (30). Briefly, lungs were harvested and dipped in 75% ethanol to devitalize the surface mesothelium. Subsequently, the lungs were minced, digested, filtered (100-μm nylon mesh), and the cell suspension was incubated with a rat anti-mouse CD31 Ab (BD Biosciences). Subsequently, the suspension was washed twice and incubated with sheep anti-rat IgG-coated micromagnetic beads (Dynal Biotech) in M199. The microbeads (with attached cells) were captured by a Dynal magnet, washed three times, and seeded in a fibronectin-coated T25 cell culture flask in EBM-2 (endothelial cell basal medium-2) media supplemented with EGM-2 (endothelial growth medium-2) MV (Cambrex), 100 U/ml penicillin G, 100 μg/ml streptomycin, and 20% FCS and cultured under standard conditions. This procedure yielded PEC monolayers with 90% homogeneity. First through third passage cells were used for experiments.
AMφ.
AMφ were obtained from bronchoalveolar lavage fluid as previously described (14). The lungs were lavaged three times with 1 ml PBS containing 2 mM EDTA. This procedure yielded 2.5 × 105 cells per mouse, 95% of which were AMφ (Wright-Giemsa). Approximately 1 × 105 cells were resuspended in MEM, seeded into each well, and allowed to attach for 30 min. After a wash, the attached AMφ were used in experiments immediately.
PMN.
PMN were obtained from the bone marrow of the hind legs of adult mice using Percoll sedimentation and gradient centrifugation as previously described (30, 34). This procedure yields 5–6 million white blood cells, 95% of which are mature PMNs (crystal violet staining). The PMN were used in experiments immediately.
Assays
PMN accumulation in pulmonary alveoli.
Lungs were harvested from anesthetized mice, perfused with 10% formalin (to inflation), and fixed for 24 h. The specimens were processed for paraffin embedding. After deparaffinization, rehydration, and Ag retrieval with sodium citrate (pH 6.0), tissue sections were probed with NIMP-R14 IgG2b Ab (Hycult Biotechnology) targeting PMN. The PMN in alveoli were identified using an anti-Ig HRP detection kit (BD Pharmingen) according to the manufacturer’s directions. The numbers of PMN in alveoli were assessed by immunofluorescence microscopy (Zeiss Axiovert 200M).
Pulmonary microvascular protein leak.
Thirty minutes before harvesting the lungs, Evans blue (EB) dye (50 mg/kg) was injected i.v. After flushing the pulmonary circulation with PBS, the lungs were removed, weighed, and homogenized in PBS. EB was extracted into formamide by incubating at 60°C for 16 h. After centrifugation, absorbance of the supernatant was measured at 620 and 740 nm (with correction for heme pigments) and quantitated using an EB standard curve (2, 11, 12, 13, 14, 35).
PMN transendothelial migration.
PEC were grown to confluence on fibronectin-coated cell culture inserts (3-μm-diameter pores; BD Biosciences). 51Cr-labeled mouse PMN were added to the apical aspect of the PEC monolayer and AMφ were added to the basal compartment. After 90 min the percentage of added PMN that migrated from the apical to the basal aspects of the insert was calculated (30). Since this assay involved the simultaneous use of isolated cells (and associated inherent variability), the same isolates were used for a given set of experiments.
Oxidant stress.
Oxidation of dihydrorhodamine 123 (DHR 123; Molecular Probes) was used as a general indicator of cellular oxidant stress (30, 36). The cells were pretreated with 20 μM DHR 123 for 1 h. After a given challenge, the cells were washed, lysed, and sonicated. The cell lysates were centrifuged and analyzed for DHR 123 oxidation at excitation and emission wavelengths of 502 and 523 nm, respectively.
Reduction of NBT to blue formazan was used as a general indicator of PEC superoxide production (37). Briefly, cells were preloaded with NBT (1 mg/ml) for 1 h before challenge. After challenge, the cells were washed with PBS and rinsed with 2 N NaOH. The insoluble formazan was dissolved in 1 ml of DMSO and OD measured at 654 nm wavelength.
Lucigenin ECL using NADPH as a substrate was used as an index of superoxide production by PEC NADPH oxidase (38). After challenge, the cells were scraped from the culture plates, centrifuged, and the cell pellet was resuspended in Kreb/HEPES buffer. An aliquot (100 μl) was mixed with 5 μl lucigenin and 0.1 mM NADPH. The luminescence signal was measured continuously for 5 min.
Western blot.
Phosphorylated and total protein of interest in endothelial cells was assessed by Western blot as previously described (39, 40). In brief, after challenge, the cells were harvested in lysis buffer, sonicated, and 5 μg of cell protein was resolved by 10% SDS-PAGE. Protein was transferred to polyvinylidene difluoride membranes, blocked, and probed with relevant Abs. After washing, the blots were incubated with HRP-conjugated secondary Abs and detected using the ECL Plus detection system (Amersham Biosciences).
ELISA.
KC (keratinocyte-derived chemokine), LIX (LPS-inducible CXC chemokine), and TNF-α levels in septic plasma in the absence and presence of AMφ were determined by ELISA (34
Knock down of c-Src kinase.
A siRNA against c-Src kinase (Santa Cruz Biotechnology) was used to knock down c-Src kinase expression in PEC as previously described (40). Transfecton was performed using TransMessenger transfection reagent (Qiagen) according to the manufacturer’s directions. The cells were used for experiments 48 h after transfection. Transfection efficiency was quantitiated by Western blot.
Experimental reagents
phox (Upstate Biotechnology), p47phox (Upstate Biotechnology), phospho-Src family (Tyr416) (Cell Signaling Technology), and Src (36D10) (Santa Cruz Biotechnology). Catalase, SOD, PAF (C18), H2O2, and apocynin were from Sigma-Aldrich. The PAF receptor antagonist (WEB2086) was from Boehringer Ingelheim. Diphenyleneiodonium and 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP2) were from Calbiochem.
Statistics
All values are expressed as means ± SEM. Statistical analysis was performed using ANOVA and a Student’s t test with Bonferroni corrections for multiple comparisons. A value of p < 0.05 was considered to be significant.
Results
Peritonitis promotes PMN infiltration of the lungs and activates AMφ: role of circulating chemokines
Introduction of FIP resulted in a progressive accumulation of PMN in the pulmonary alveolar space at 6 and 12 h after FIP (Fig. 1⇓A). As shown in Fig. 1⇓B, FIP also resulted in a progressive increase in pulmonary vascular protein leak over the same timeframe. Collectively, these findings are in agreement with previous studies indicating that sepsis initiated in the peritoneum (by CLP or endotoxin) can promote PMN infiltration of the pulmonary alveolar space and vascular protein leak within 2–6 h after challenge (4, 23, 24, 25, 26, 27, 41). Mortality in our experimental model of peritonitis was negligible until 6–8 h after introduction of FIP, at which point it increased to 5% and by 12 h it was 20% (n = 21). Thus, for further studies, we chose the 6 h time point to address the early pulmonary inflammation in the FIP model.
Sepsis induced by FIP results in pulmonary inflammation. A, FIP-induced PMN accumulation in alveoli. Representative immunohistochemistry (above) and quantitation of the number (below) of PMN in the alveolar space at 6 and 12 h after FIP. The results presented are the average PMN per alveoli as a function of lung surface area (mm2). Data are from five high-power fields from two to three slides from each of three mice. B, Pulmonary vascular leak of EB-protein at 6 and 12 h after FIP. n = 3; *, p < 0.05 compared with sham.
Plasma obtained from mice 6 h after FIP was able to activate naive AMφ, as evidenced by an increase in AMφ oxidative stress (Fig. 2⇓). An immunoneutralization approach was used to assess whether circulating chemokines could activate AMφ. Abs directed to chemokines (KC or LIX) prevented AMφ activation by septic plasma (Fig. 2⇓A). As another approach to address this issue, AMφ isolated from CXCR2−/− mice were used. The CXCR2−/− AMφ did not incur an oxidant stress when exposed to septic plasma (Fig. 2⇓B). Thus, taken together, these findings indicate that chemokines present in septic plasma can activate AMφ.
Chemokines in plasma from septic mice increase oxidant stress (DHR oxidation) in naive AMφ. A, Naive AMφ, attached to culture plates, were loaded with DHR 123. Plasma was obtained from mice 6 h after FIP-induced peritonitis, diluted to 20%, and added to AMφ. One hour later DHR 123 oxidation was assessed. Abs against LIX or KC (1.0 μg/ml) were added to plasma 15 min before challenge of AMφ. n = 3; *, p < 0.05 vs sham plasma. B, Naive AMφ from WT or CXCR2−/− C57BL/6-BALB/c hybrid mice were subjected to the same protocol as in A. n = 3; *, p < 0.05 as compared with sham in appropriate group.
AMφ can promote PMN transendothelial migration against a concentration gradient induced by septic plasma
During FIP-induced sepsis, the circulating blood would be expected to contain higher concentrations of inflammatory mediators (e.g., chemokines) than the pulmonary interstitial fluid. To mimic this situation in vitro, the concentrations of septic plasma on the apical (vascular) and basal (interstitial) aspects of the PEC monolayers were altered accordingly and PMN transendothelial migration was assessed. As shown in Fig. 3⇓A (first and second bars), PMN transendothelial migration was reduced when the concentration of septic plasma was lower in the basal compartment than in the apical one. This observation is consistent with the present dogma that a concentration gradient of inflammatory mediators (e.g., chomkines) determines the direction of PMN migration (2, 5, 6, 7, 8, 9).
AMφ can promote PMN transendothelial migration against a concentration gradient established by septic plasma. A, Naive AMφ were allowed to attach to the basal aspect of the cell culture inserts. Septic plasma was obtained 6 h after FIP-induced peritonitis, diluted to 10% or 20%, and added to the basal aspects of the inserts. Immediately thereafter, PMN suspended in 20% plasma were added to the apical aspects of the inserts. Ninety minutes later PMN migration was assessed. n = 3; *, p < 0.05 compared with first bar; #, p < 0.05 compared with the relevant conditions without AMφ. B, Naive AMφ were incubated with 20% septic plasma for 90 min and the plasma concentrations of KC, LIX, and TNF-α were assessed (ELISA). As a control AMφ were incubated with MEM. n = 3; *, p < 0.05 as compared with AMφ in MEM.
To determine whether AMφ could promote PMN transendothelial migration under septic conditions, a similar approach was used. As shown in Fig. 3⇑A (first and third bars), when naive AMφ were placed in the basal compartment of the cell culture inserts with an equal concentration of septic plasma on the apical and basal aspects of the PEC monolayers, they were able to promote PMN transendothelial migration. More importantly, when the apical aspect of the PEC monolayers contained twice the concentration of septic plasma as the basal aspect, the AMφ were still capable of promoting PMN transendothelial migration (Fig. 3⇑A, second and fourth bars). These findings indicate that AMφ can promote PMN transendothelial migration even against a concentration gradient induced by septic plasma.
As shown in Fig. 3⇑B, septic plasma contains the cytokine, TNF-α, and the chemokines LIX and KC. Coincubation of AMφ with septic plasma for 90 min (duration of the migration assay in Fig. 3⇑A) did not result in any increment in the concentrations of TNF-α, LIX, or KC. These findings indicate that the ability of AMφ to promote PMN migration against a concentration gradient established by septic plasma is not due to AMφ production of these inflammatory mediators.
AMφ from septic mice can promote PMN transendothelial migration: role of PAF and H2O2
As another approach to studying the role of AMφ in promoting PMN transendothelial migration, AMφ were harvested from bronchoalveolar lavage fluid obtained 6 h after FIP. As shown in Fig. 4⇓, AMφ from septic mice, in the absence of any further stimulation, promoted PMN transendothelial migration. These findings support the contention that AMφ can be activated during sepsis initiated at a remote locus (e.g., peritoneum) presumably by circulating inflammatory mediators, such as chemokines (Fig. 2⇑). This latter approach entailing the use of AMφ from septic mice (septic AMφ), rather than the use of AMφ from naive mice challenged with septic plasma, was chosen for further study.
AMφ from septic mice promote PMN transendothelial migration by generating PAF and H2O2. Six hours after FIP-induced peritonitis, AMφ were harvested from septic animals, allowed to attach to the basal aspect of the cell culture inserts, and PMN transendothelial migration was assessed 90 min later. A, Effects of mAbs against LIX or KC (1 μg/ml). B, Effects of the PAF antagonist (WEB2086; 20 μg/ml). C, Effects of catalase (2000 U/ml) or SOD (300 U/ml). Abs, antagonist, and antioxidant enzymes were added to the basal compartment 15 min before beginning the migration assay. n = 3; *, p < 0.05 as compared with sham; #, p < 0.05 as compared with FIP.
Chemokines, PAF, and reactive oxygen metabolites are produced by activated AMφ and have been implicated in the sepsis-induced pulmonary inflammation (1, 42). As shown in Fig. 4⇑A, the PMN transendothelial migration induced by septic AMφ was not affected by Abs directed to the chemokines LIX and KC. A PAF receptor antagonist (WEB2086) completely prevented the PMN transendothelial migration induced by septic AMφ (Fig. 4⇑B). Interestingly, the PMN transendothelial migration induced by AMφ derived from septic mice was not affected by SOD, but it was completely prevented by catalase (Fig. 4⇑C). Collectively, these findings indicate that both PAF and H2O2, but not LIX or KC, contribute to the PMN migration induced by septic AMφ.
To gain further insight into the mechanisms involved in PMN transendothelial migration induced by septic AMφ, the effects of catalase, SOD, and the PAF antagonist were evaluated using acellular filters (no endothelium). Only the PAF antagonist was capable of preventing PMN migration across acellular filters (data not shown). The effect of the PAF antagonist reflects the ability of AMφ-generated PAF to activate and attract PMN. However, in the complete system it is more likely that the PAF antagonist is exerting its effect on the endothelium, rather than by penetrating the endothelial monolayer to affect PMN in the apical compartment. This, coupled to the lack of effect of catalase in the absence of endothelium, suggests that the endothelium may play an important role the PMN transendothelial migration induced by septic AMφ.
Endothelial cell oxidant stress induced by septic AMφ is a prerequisite for PMN transendothelial migration: role of PAF and H2O2
Evidence is accumulating to support the contention that endothelial cell activation plays an important role in PMN transendothelial cell migration (28, 30, 42, 43). As shown in Fig. 5⇓, A and B, supernatants from septic AMφ can activate PEC, as evidenced by an increase in PEC oxidant stress. The PAF receptor antagonist (WEB2086) prevented the AMφ-induced increase in PEC oxidative stress (Fig. 5⇓A). Furthermore, catalase, but not SOD, prevented the increase in PEC oxidative stress (Fig. 5⇓B). These observations indicate that both PAF and H2O2 are required for PEC activation by septic AMφ. To assess whether PAF and H2O2 could act in concert to induce an oxidant stress in PEC, they were used individually or in combination. As shown in Fig. 5⇓C, PAF and H2O2, when used individually, did not significantly induce an oxidative stress in PEC. However, when used in combination, PEC oxidant stress was increased. This latter observation provides a potential explanation for why blockade of either H2O2 or PAF can completely prevent PEC oxidant stress induced by septic AMφ; that is, since both are required, blockade of either would prevent PEC oxidative stress.
PEC oxidative stress (DHR oxidation) induced by AMφ from septic mice is due to PAF and H2O2. Six hours after FIP-induced peritonitis, AMφ were harvested and allowed to attach to cell culture plates. The AMφ were washed, incubated with MEM for 1 h, and the supernatants were collected and added to PEC monolayers (preloaded with DHR 123). One hour later DHR oxidation was assessed. A, The PAF antagonist (WEB2086; 20 μg/ml) was added to the supernatants 15 min before challenge of PEC. B, Catalase (2000 U/ml) or SOD (300 U/ml) was added to the supernatants 15 min before challenge of PEC. C, PAF (10−9 M) and H2O2 (200 μM) were added to PEC monolayers and 1 h later DHR oxidation was assessed. n = 3; *, p < 0.05 as compared with sham (or MEM); #, p < 0.05 as compared with FIP.
To assess whether the PEC oxidant stress was causally linked to PMN transendothelial migration induced by septic AMφ, we used PEC isolated from Cu/Zn SOD overexpressing mice (Cu/Zn SOD Tg). As shown in Fig. 6⇓A, the increased oxidant stress in PEC induced by septic AMφ was abolished when PEC derived from Cu/Zn SOD mice were used. Furthermore, septic AMφ did not promote PMN migration across monolayers of PEC that were overexpressing Cu/Zn SOD (Fig. 6⇓B). These observations suggest that PEC superoxide plays an important role in the PMN transendothelial migration induced by septic AMφ.
Overexpression of Cu/Zn SOD in PEC abrogates the AMφ-induced PEC oxidant stress (DHR oxidation) and PMN transendothelial migration. PEC were isolated from Cu/Zn SOD overexpressing mice (Cu/Zn SOD Tg) or their WT counterparts. Six hours after FIP-induced peritonitis, AMφ were harvested from septic mice and allowed to attach to cell culture plates. A, The AMφ were incubated with MEM for 1 h and the supernatants were collected, added to PEC monolayers, and DHR oxidation was assessed 1 h later. B, AMφ-induced PMN transendothelial migration was assessed 90 min later as described in Fig. 4⇑. n = 3; *, p < 0.05 compared with sham within the appropriate group.
Role of PEC NADPH oxidase in the PMN transendothelial migration induced by septic AMφ
Since NADPH oxidase is a major intracellular source of superoxide in endothelial cells (29, 31, 32, 44), we assessed the role of PEC NADPH oxidase in the endothelial cell-dependent PMN migration induced by septic AMφ. As shown in Fig. 7⇓A, both a membrane-associated subunit of NADPH oxidase (gp91phox) and a cytosolic subunit (p47phox) were present in PEC. Supernatants from septic AMφ did not affect the protein levels of either gp91phox or p47phox (Fig. 7⇓A). However, the supernatants from septic AMφ increased NADPH oxidase activity, as evidenced by NADPH-dependent superoxide generation (Fig. 7⇓B). These findings indicate that septic AMφ activate PEC NADPH oxidase via nongenomic signaling.
AMφ from septic mice increase NADPH oxidase activity in PEC. Six hours after FIP-induced peritonitis, AMφ were harvested from septic mice and allowed to attach to cell culture plates. The AMφ were washed, incubated with MEM for 1 h, and the supernatants were collected and added to PEC monolayers for 1 h. A, Representative Western blots (of three experiments) indicating that PEC gp91 and p47 protein expression is not altered by supernatants from septic AMφ. B, NADPH oxidase activity (lucigenin ECL) in PEC monolayers exposed to supernatants from septic AMφ for 1 h. n = 3; *, p < 0.05 compared with sham.
Two approaches were used to inhibit NADPH oxidase (i.e., pharmacologic and genetic). Both diphenyleneiodonium and apocynin inhibited both PEC superoxide production and PMN transendothelial migration induced by septic AMφ (data not shown). However, since both of these pharmacologic inhibitors may be rather nonspecific (38), we also made use of PEC isolated from gp91phox-deficient (gp91−/−) mice. Both the PEC superoxide production (Fig. 8⇓, A and B) and PMN migration (Fig. 8⇓C) were abolished when gp91−/− PEC were used in the assays. Taken together, these observations indicate that PEC NADPH oxidase-derived superoxide plays a critical role in the PMN migration induced by septic AMφ.
The increase in PEC NADPH oxidase activity, superoxide production, and PMN transendothelial migration induced by septic AMφ is abrogated when gp91-deficient PEC are used in the assays. PEC were isolated from gp91phox-deficient mice (gp91−/−) or their WT counterparts. Six hours after FIP-induced peritonitis, AMφ were harvested from septic mice and allowed to attach to cell culture plates. A and B, The AMφ were washed, incubated with MEM for 1 h, and the supernatants were collected and used to measure PEC NADPH oxidase activity (lucigenin ECL) (A) and superoxide production (NBT reduction) (B). C, PMN transendothelial migration was assessed as described in Fig. 4⇑. n = 3; p < 0.05 compared with sham in relevant group.
PMN transendothelial migration induced by septic AMφ involves Src kinase upstream of NADPH oxidase in PEC
Previous studies indicate that Src kinase activation may be both an upstream and downstream event of NADPH oxidase activation (45, 46). As shown in Fig. 9⇓A, phosphorylation of PEC Src family kinase (Tyr416; activation loop) was increased within 1 min after addition of supernatants from septic AMφ. The septic AMφ-induced phosphorylation of Src kinase was inhibited by both catalase and the PAF receptor antagonist (Fig. 9⇓B).
PEC Src kinase is phosphorylated by septic AMφ. Six hours after FIP-induced peritonitis, AMφ were harvested from septic mice and allowed to attach to cell culture plates. The AMφ were washed, incubated with MEM for 1 h, and the supernatants were collected and used to measure PEC Src kinase phosphorylation (Western blots). A, Time course of Src kinase phosphorylation. n = 3; *, p < 0.05 compared with the 0 time point. B, The PAF antagonist (WEB2086; 20 μg/ml) or catalase (2000 U/ml) were added to the supernatants 15 min before challenge of PEC. Src kinase phosphorylation was assessed 5 min after challenge of PEC. n = 3; *, p < 0.05 as compared with sham; #, p < 0.05 as compared with FIP.
In PEC challenged with supernatants from naive (sham) AMφ (Fig. 10⇓), p47phox was localized largely to the perinuclear region. Within 10 min after challenge of PEC with supernatants from septic AMφ, p47phox was noted throughout the cytoplasm. Two approaches were used to block Src kinase: pharmacologic inhibition of Src family kinases with PP2 (47), and a knock down of Src kinase with siRNA. PP2 completely prevented the redistribution of intracellular p47phox from the perinuclear region to the cell periphery (Fig. 10⇓A). PP2 is nonspecific and can inhibit other members of the Src family (e.g., Lck) (47). Transfection of PEC with Src siRNA reduced Src protein levels by ∼50% and reduced the effect of septic AMφ on redistribution of p47phox (Fig. 10⇓B).
PEC Src kinase plays a role in the redistribution of intracellular p47phox induced by septic AMφ. Six hours after FIP-induced peritonitis, AMφ were harvested from septic mice and allowed to attach to cell culture plates. The AMφ were washed, incubated with MEM for 1 h, and the supernatants were collected and added to PEC grown on coverslips. After a 10-min incubation, PEC were washed, fixed, and probed with rabbit anti-mouse p47phox (Upstate Biotechnology) and stained with Texas Red-labeled secondary Ab (goat anti-rabbit; Invitrogen). PEC nuclei were stained with Hoechst. In some experiments, PEC were pretreated with the Src kinase inhibitor PP2 1 h before the experiments (A) or transfected with siRNA targeting Src kinase 48 h before the experiments (B). The cells were examined under 60× oil objective by immunofluorescence microscopy (Zeiss Axiovert 200M). Representative fluorescent images are from three experiments. Lower panel in B is a representative Western blot depicting PEC Src kinase levels 48 h after transfection.
PP2 also prevented the increase in PEC NADPH oxidase activity and PMN transendothelial migration induced by septic AMφ (Fig. 11⇓A). When these experiments were repeated with PEC transfected with Src siRNA, there was a 50% reduction in NADPH oxidase activity and a 35% reduction in PMN transendothelial migration (Fig. 11⇓B). The reasons for the partial inhibition by the siRNA can be attributed to either 1) the limited transfection efficiency (50%) and/or 2) involvement of other members of the Src family of kinases. The siRNA data favor a role for PEC Src kinase in the activation of NADPH oxidase activity and the resultant AMφ-induced PMN transendothelial migration. However, we cannot exclude a role for other Src family members in NADPH oxidase activation.
PEC Src kinase plays a role in the increase in PEC NADPH oxidase activity and PMN transendothelial migration induced by septic AMφ. Six hours after FIP-induced peritonitis, AMφ were harvested and allowed to attach to cell culture plates. NADPH oxidase activity (chemiluminescence) was measured as described in Fig. 8⇑. PMN transendothelial migration was assessed as described in Fig. 4⇑. PEC were pretreated with the Src kinase inhibitor PP2 1 h before the experiments (A) or transfected with siRNA targeting Src kinase 48 h before the experiments (B). n = 3; *, p < 0.05 compared with sham AMφ; #, p < 0.05 compared with FIP AMφ.
Discussion
A common pathophysiological feature of ALI/ARDS induced by abdominal sepsis (or peritonitis) is PMN infiltration of the lungs and vascular protein leak (4, 25, 26, 27, 41). Pulmonary PMN infiltration would be facilitated by the establishment of an interstitial-to-blood chemotactic gradient by resident macrophages (1, 2). In this scenario, cytokines generated within the infected peritoneum enter the systemic circulation and activate AMφ, which in turn generate chemokines to promote PMN infiltration of the lung (12, 13, 48, 49). However, the validity of this scenario has not been experimentally assessed. Thus, one of the major objectives of the present study was to determine whether AMφ are activated during sepsis and whether these activated AMφ can promote PMN transendothelial migration. Using a peritonitis model of sepsis and a reductionist approach to mimic the vascular-interstitial interface, we herein provide the following novel insights into the role of AMφ in PMN emigration into the lungs during peritonitis. First, AMφ obtained from mice rendered septic by FIP could promote PMN transendothelial migration in the absence of any further stimulation (Fig. 4⇑). Second, chemokines in septic plasma can activate AMφ via ligation of CXCR2 (Fig. 2⇑). Third, the activated AMφ can promote PMN migration against a concentration gradient established by septic plasma (Fig. 3⇑). Finally, AMφ-derived H2O2 and PAF, but not chemokines, mediate the AMφ-induced PMN transendothelial migration (Fig. 4⇑). Collectively, these observations argue against the contention that cytokines released into the circulation during peritonitis activate AMφ that promote PMM emigration via release of chemokines.
Using immunoneutralization (Abs) and genetic blockade (CXCR2−/− AMφ) approaches, we show that the ability of septic plasma to activate AMφ can be attributed to the presence of chemokines therein (Fig. 2⇑). In general, chemokine activation of AMφ is dependent on CCR, while chemokine activation of PMN is dependent on CXCR (50). However, both AMφ and PMN express the two chemokine receptors (51), albeit CXCR2 levels are higher on PMN, while CCR2 levels are higher on monocytic cells. Although CXCR2 have been implicated in PBMC migration (51), to our knowledge our observations represent the first direct evidence that AMφ CXCR2 occupancy is critical to AMφ activation and their ability to promote PMN transendothelial migration during peritonitis-induced sepsis.
The present consensus holds that in ALI/ARDS the activated AMφ generate chemokines that facilitate PMN infiltration of the lungs (2, 12, 13, 16, 48). During peritonitis-induced sepsis the circulating blood would be expected to contain higher concentrations of inflammatory mediators (e.g., chemokines) than the pulmonary interstitial fluid, at least initially. The observation that AMφ can promote PMN transendothelial migration against a concentration gradient of septic plasma (Fig. 3⇑A) suggests that they must either produce chemokines/cytokines in amounts to exceed the plasma concentration or that they are generating other inflammatory mediators to overcome the plasma-to-interstitial chemotactic gradient (6, 52). Our findings (Figs. 3⇑B and 4⇑) support the latter possibility; that is, AMφ obtained from septic mice promote PMN migration by generating PAF and H2O2, rather than chemokines/cytokines.
The scenario uncovered in the present study using an in vitro construct of the pulmonary vascular-interstitial interface is markedly different from our previous in vitro studies of the myocardial vascular-interstitial interface during peritonitis (34). In the latter model, the effector cell was the cardiac myocyte, rather than AMφ. Activation of cardiac myocytes with septic plasma converted them to a proinflammatory phenotype; that is, these myocytes incurred an oxidant stress and promoted PMN transendothelial migration. However, cardiomyocyte activation by septic plasma involved the cytokines TNF-α and IL-1β, whereas in the present study AMφ activation by septic plasma could be entirely accounted for by chemokines (Fig. 2⇑). Furthermore, the myocyte-induced PMN migration was completely prevented by Abs directed to chemokines, while in the present study the AMφ-induced PMN migration was not affected by immunoneutralization of the chemokines (Fig. 4⇑). Thus, during sepsis, the inflammatory mediators that both activate interstitial effector cells and contribute to effector cell generation of a chemotactic gradient may vary substantially from one organ system to another. This variability may in part explain the failure of clinical trials evaluating the protective effect of a single inflammatory mediator in septic patients (1, 3, 10).
Another major objective of the present study was to address the role of endothelial cell oxidant stress in the PMN transendothelial migration induced by septic AMφ. Herein, we provide evidence that septic AMφ can induce an oxidant stress in endothelial cells, an effect attributed to both H2O2 and PAF (Fig. 5⇑). Although it seems logical that H2O2 would induce an oxidative stress in PEC, it is not readily apparent how PAF induces an oxidant stress. Previous studies indicate that PAF receptor antagonists can inhibit H2O2 secretion by activated monocytes/macrophages (53). Our experimental design negates this possibility, since the PAF receptor blockade was applied after the septic AMφ had generated relevant substances (Fig. 5⇑). Another possibility is that PAF and H2O2 may act synergistically. Previous studies indicate that H2O2-induced neuronal injury (apoptosis) can be potentiated by PAF (54). Our observations support this latter possibility, since PAF and H2O2 acted synergistically to increase PEC oxidant stress (Fig. 5⇑C).
Our previous studies indicate that endothelial cell generation of superoxide is a prerequisite for the PMN migration induced by cardiac myocytes conditioned with anoxia/reoxygenation (30). A similar situation appears to hold in our sepsis model; that is, both the oxidant stress in PEC and the PMN transendothelial migration induced by septic AMφ did not occur when PEC isolated from Cu/Zn SOD overexpressing mice were used (Fig. 6⇑). These observations would indicate that, while H2O2 may be the AMφ-derived oxidant that initially impacts PEC, superoxide is somehow involved in the signaling cascade within endothelium that ultimately contributes to PMN transendothelial migration. Further studies indicated that the H2O2-driven source of superoxide in PEC is NADPH oxidase (Figs. 7⇑ and 8⇑).
NADPH oxidase activation involves the activation of cytoplamic subunits (e.g., p47phox) and their mobilization and merger with the membrane subunits (e.g., gp91phox). Considering the limited diffusion of H2O2 within the cytoplasm, it seemed unlikely that AMφ-derived H2O2 would be directly interacting with p47phox. Thus, we considered the possibility that some membrane-associated signaling kinase may be responsible for NADPH oxidase activation in our model. In this regard, c-Src kinase appeared to be a likely candidate since 1) its signaling activity is dependent on membrane association (55), 2) it can be activated by both PAF and H2O2 (56), and 3) it has been proposed to be involved in NADPH oxidase signaling as both an upstream and downstream component (31, 46). Herein, we show that PEC Src family kinase activation (phosphorylation) is increased by supernatants from septic AMφ, an event attributed to H2O2 and PAF (Fig. 9⇑). Additionally, an inhibitor of Src family kinase completely prevented the septic AMφ-induced NADPH oxidase activation as evidenced by redistribution of p47phox from perinuclear to peripheral regions of PEC (Fig. 10⇑A). Knockdown of Src kinase was only partially effective (Fig. 10⇑B). Similarly, the septic AMφ-induced PEC superoxide production and PMN transendothelial migration were completely prevented by the inhibitor of Src family kinase (Fig. 11⇑A), while knockdown of Src kinase was only partially effective (Fig. 11⇑B). Thus, although our siRNA data support a role for Src kinase in the activation of NADPH oxidase in our model, we cannot exclude the participation of other members of the Src kinase family.
Collectively, our findings indicate that the endothelial cell-dependent PMN migration involves a PAF/H2O2-Src kinase/NADPH oxidase pathway. Previous studies have implicated various individual components of this pathway in endothelial cell oxidant stress and/or monocyte adhesive interactions with endothelium induced by pharmacologic agents (e.g., LPS, PAF, and H2O2) (33, 46, 57, 58, 59). The significance of our findings is 1) the use of an in vitro construct of the pulmonary vascular-interstitial interface using cells relevant to sepsis-induced ALI/ARDS (e.g., AMφ), and 2) the sequential positioning of these components into a pathway responsible for endothelial-dependent PMN migration.
The mechanisms by which the Src kinase/NADPH oxidase pathway facilitates PMN transendothelial migration are not clear. Assuming that PMN transendothelial migration in vitro occurs primarily via the paracellular pathway, then PMN emigration would be facilitated by endothelial cell retraction and/or disruption of junctional complexes, events intimately linked to cytoskeleton function. Interestingly, there are several lines of evidence supporting a potential role for the Src kinase/NADPH oxidase pathway in regulating cytoskeleton function and/or junctional integrity in endothelial cells. First, both Src kinase and p47phox appear to interact with an actin-binding protein important for PMN transendothelial migration (43, 60, 61). Second, both Src kinase and NADPH oxidase have been implicated in increases in endothelial monolayer permeability in vivo and in vitro (24, 62, 63). Although largely circumstantial, collectively the available information warrants a systematic assessment of the downstream signaling components and mechanism of the Src kinase/NADPH oxidase regulation of endothelial barrier integrity.
In the present study, we focused on the mechanisms involved in PMN transendothelial migration. However, to reach the alveolar space (Fig. 1⇑), circulating PMN must cross two cellular barriers: the capillary endothelium and pneumocytes (type I or type II). While not addressed in the present study, others have assessed PMN migration across 549 cells (human pulmonary type II-like epithelial cells). Stimulation of A549 cell monolayers with either TNF-α or Pseudomonas aeruginosa enhanced PMN transepithelial migration (64, 65). While these cells produced IL-8, this chemokine was not the major mediator of PMN migration. One candidate mediator is the eicosanoid hepoxilin A3 (65). If one considers the two barriers in tandem, the situation appears to be even more complex. Studies on endothelial-epithelial bilayers indicate that LPS-induced PMN transendothelial migration can be inhibited by the presence of an adjacent A549 epithelial monolayer (66). Thus, the situation in vivo is most likely more complex then that simulated by our in vitro construct, and further studies are warranted to more clearly define the cellular mechanisms involved in PMN emigration from the circulation into the alveolar space.
In summary, peritonitis leads to activation of AMφ and PMN emigration into pulmonary alveoli. Using an in vitro construct of the pulmonary vascular-interstitial interface we provide the following insights into the role of AMφ in PMN transendothelial migration during peritonitis. First, CXC chemokines in septic plasma can activate AMφ that are capable of promoting PMN transendothelial migration against a chemotactic gradient established by septic plasma. Second, septic AMφ promote PMN migration by generating PAF and H2O2. Third, PAF and H2O2 induce an oxidant stress in endothelial cells that appears to be a prerequisite for the AMφ-induced PMN transendothelial migration. Finally, the intracellular Src kinase/NADPH oxidase pathway in endothelial cells plays an important role in the PMN transendothelial migration induced by septic AMφ.
Disclosures
The authors have no financial conflicts of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by Canadian Institutes for Health Research Grants MOP-37758, MGC-12816, and MOP-81303.
↵2 Address correspondence and reprint requests to Dr. Peter Kvietys, Critical Illness Research, Lawson Health Research Institute, London Health Sciences Centre, WC, 800 Commissioners Road E, London, Ontario N6A 4G4, Canada. E-mail address: peter.kvietys{at}lhsc.on.ca
↵3 Abbreviations used in this paper: ALI, acute lung injury; AMφ, alveolar macrophage; ARDS, acute respiratory distress syndrome; CLP, cecal ligation and perforation; DHR, dihydrorhodamine; EB, Evans blue; FIP, feces in the peritoneum; KC, keratinocyte-derived chemokine; LIX, LPS-inducible CXC chemokine; PAF, platelet-activating factor; PEC, pulmonary endothelial cell; PMN, polymorphonuclear leukocyte; PP2, 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine; siRNA, small interfering RNA; SOD, superoxide dismutase; WT, wild type.
- Received May 12, 2008.
- Accepted October 14, 2008.
- Copyright © 2008 by The American Association of Immunologists
References
In this issue
