METHODS

Dissection and incubation of pulmonary arteries

Lungs were obtained from White Landrace male pigs of 20–25 kg body weight. All animal experiments were conducted in accordance with the rules and regulations of Bristol University and the Home Office regarding the care and use of experimental animals. Pigs were anaesthetised with an intravenous injection of ketamine hydrochloride (10 mg/kg; Ketaset Injection, Fort Dodge Animal Health, Southampton, UK) and inhaled oxygenated halothane. The internal carotid artery was exposed, a cannula was placed in the carotid artery, and the animals were then exsanguinated. The chest was opened by median sternotomy and the lungs excised from the chest. Pulmonary arteries (PA) of 3–4 mm diameter (1st order) or 200–400 μm (4th order) were dissected from the lungs within 30 minutes and placed in Dulbecco’s Minimum Essential Medium supplemented with Glutamax-1, 100 U/ml penicillin, and 100 μg/ml streptomycin (DMEM; GibcoBRL, Paisley, UK). The arteries were cut into 2–3 mm² segments for experimentation. In some studies the endothelium was removed by gently rubbing the luminal surface with a cotton wool bud.

The segments were then incubated for up to 16 hours in serum-free DMEM containing LPS (E coli 026:B6; Sigma, Poole, Dorset, UK), human recombinant IL-1α (R&D Systems, Abingdon, UK), and human recombinant TNF-α (R&D Systems) alone and in combination with each other. After incubation the segments were washed in Dulbecco’s phosphate buffered saline (PBS, GibcoBRL) and the formation of O₂^{• −} was measured.

Measurement of superoxide (O₂^{• −})

The measurement of O₂^{• −} release by arterial segments was performed by detection of ferricytochrome c reduction.¹³ Following incubation, arterial segments were rinsed three times with PBS and equilibrated in DMEM without phenol red for 10 minutes at 37°C in a 95% air/5% CO₂ incubator (Heraeus, Hera Cell, Kandro Laboratory Products, Germany). 20 μM horseradish cytochrome c (Sigma) with or without 500 U/ml copper-zinc superoxide dismutase (SOD; Sigma) was added to the segments and incubated at 37°C in a 95% air/5% CO₂ incubator for 1 hour. The final volume of the reaction mixture was 0.5 ml/well. After 1 hour the reaction medium was removed and the maximum rate of reduction of cytochrome c was determined at 550 nm on a temperature controlled anthos Lucy 1 spectrometer (Lab-tech International, Ringmer, East Sussex, UK) and converted to nmol O₂^{• −} using ΔE_{550 nm} = 21.1/mM/cm as the extinction coefficient for (reduced-oxidised) cytochrome c. The reduction of cytochrome c that was inhibitable with SOD reflected actual O₂^{• −} release. Segments were blotted, dried and weighed, and the data were expressed as nmol O₂^{• −} /mg tissue/h.

Effect of enzyme and NOS inhibitors on O₂^{• −} release

To determine the source of the O₂^{• −}, PA segments were preincubated with 10 μM diphenylene iodonium chloride (DPI, Sigma), an NADPH oxidase inhibitor; 10 μM rotenone (Sigma), an inhibitor of mitochondrial respiration; and 100 μM allopurinol (Sigma), an inhibitor of xanthine oxidase, for 2 hours before measurement of O₂^{• −}.

To study a possible role for NOS derived NO in modifying O₂^{• −} formation (O₂^{• −} + NO = ONOO⁻), the effect of NOS inhibitors was studied using (1) the non-specific NOS inhibitor l-nitroarginine methyl ester (l-NAME, 100 μM; Sigma), (2) the eNOS inhibitor N⁵-(1-iminoethyl)-ornithine (l-NIO, 10 μM; Sigma), and (3) the iNOS inhibitor, l-N⁶-(1-iminoethyl)-lysine-HCl (l-NIL, 10 μM; Sigma).¹⁴ The production of O₂^{• −} was measured by SOD inhibitable reduction of ferricytochrome c as above.

Immunocytochemistry of NOS and nitrated tyrosine

Immunocytochemical analysis of eNOS, iNOS, and nitrated tyrosine (NT) was carried out in selected samples following incubation for 16 hours with LPS (1 μg/ml), IL-1α, or TNF-α (both 10 ng/ml) which were then snap frozen in liquid nitrogen and stored at −80°C. Cryostat sections (8 μm) were prepared and fixed in acetone for 10 minutes. The endogenous peroxide activity was inhibited with 1.2% H₂O₂ in methanol for 30 minutes. Sections were then treated with horse (for NT staining) or goat serum (for eNOS and iNOS) diluted 1:3 with Tris-buffered saline (TBS; Sigma), pH 7.4, drained and incubated with monoclonal antibodies against eNOS, iNOS (Transduction Laboratories, Oxford, UK) and nitrated tyrosine (Upstate Biotechnology, Buckinghamshire, UK) at a dilution of 1:200 for iNOS and eNOS and 0.3 μg/ml for NT overnight at 4°C. After washing in TBS the sections were treated with either biotinylated goat anti-rabbit (for NT; 1:200 dilution) or biotinylated goat anti-mouse (for eNOS/iNOS; 1:200 dilution) for 1 hour at room temperature, washed and further treated for an hour with avidin-biotin-peroxidase complex (Dako Ltd, Ely, Cambridgeshire, UK) as described in the manufacturer’s manual. The bound antibody was visualised by addition of 0.05% diaminobenzidine (DAB, Dako) and 0.03% hydrogen peroxide in PBS, which formed an insoluble brown precipitate (positive staining). The nuclei were counterstained using Mayer’s haematoxylin, dehydrated, and mounted. Control slides (an irrelevant isotype matched antibody in place of the primary antibody) were prepared to test the specificity of staining.

Statistical analysis

Statistical analysis was carried out using Instat (Graphpad Software Inc, San Diego, USA). Before undertaking the study, power analysis was carried out from which it was determined that an n of 6 was required for statistical assurance. A Kolmogorov-Smirnov test showed that the data were normally distributed. The data are thus expressed as mean (SE), n=6. Analysis was performed using ANOVA and a post hoc unpaired two tailed Student’s t test with Bonferonni’s adjustment.

RESULTS

LPS, IL-1α, and TNF-α promoted the formation of SOD inhibitable O₂^{• −} from porcine PA segments compared with untreated controls in a time and concentration dependent manner (figs 1 and 2). Since maximal O₂^{• −} formation was observed after 16 hours incubation, this time point was used in all subsequent studies. Similarly, the concentrations at which optimal responses were obtained were used in subsequent experiments (10 ng /ml for both IL-1α and TNF-α, and 1 μg/ml for LPS). Removal of the endothelium enhanced the formation of O₂^{• −} from porcine PA segments compared with untreated controls in response to LPS, IL-1α, and TNF-α following incubation for 16 hours (fig 2). When comparing first order arterial segments (2–4 mm diameter) with fourth order arterial segments (200–300 μm diameter), with and without endothelium, O₂^{• −} formation and release in response to LPS, IL-1α and TNF-α was similar between the two different sized vessels, although the absolute release of O₂^{• −} was greater in the smaller vessels (fig 3).

DPI, an inhibitor of NADPH oxidase (but not rotenone) inhibited the formation O₂^{• −} formation and release from porcine PA segments (with and without endothelium) in response to LPS, IL-1α, and TNF-α (fig 4). Allopurinol had a slight but still significant effect on O₂^{• −} formation induced with LPS and TNF-α but had no effect on IL-1α induced effects (fig 4).

The non-specific NOS inhibitor l-NAME enhanced O₂^{• −} formation and release from porcine PA segments with and without endothelium in response to IL-1α and TNF-α, but only when the endothelium was present in response to LPS (fig 5). The eNOS inhibitor l-NIO enhanced O₂^{• −} formation from porcine PA segments with endothelium only in response to IL-1α and TNF-α and had a lesser effect on LPS mediated O₂^{• −} formation (fig 5). The iNOS inhibitor, on the other hand, enhanced O₂^{• −} formation from porcine PA segments with endothelium only in response to LPS, but had no effect on IL-1α and TNF-α mediated O₂^{• −} formation (fig 5).

Following incubation for 16 hours the immunoreactive distribution of eNOS, iNOS, and nitrated tyrosine (index of ONOO⁻ formation) in response LPS, IL-1α, and TNF-α was assessed. Nitrated tyrosine was located principally in the endothelium (fig 6A–E), indicating that both NO and O₂^{• −} are present at high concentrations in this region. Similarly, eNOS was located in the endothelium and was markedly upregulated by IL-1α and TNF-α and, to a lesser extent, by LPS (fig 6F–J), confirming the effects of the NOS inhibitors. In contrast, iNOS was upregulated only in response to LPS and was located ubiquitously through the arterial segment, and not to IL-1α or TNF-α (fig 6K–O), again confirming the observations with the NOS inhibitors.

DISCUSSION

This study shows that TNF-α, IL-1α, and LPS promote the formation of O₂^{• −} in PA segments (with and without intact endothelium) in a time dependent manner and at concentrations that have been reported to appear in the blood of patients with ARDS.¹⁰ These responses were not confined to large pulmonary arteries but were seen also in arterial microvessels, indicating that these effects occur ubiquitously throughout the pulmonary vasculature. Furthermore, DPI completely inhibited the generation of O₂^{• −} in response to TNF-α, IL-1α and LPS, indicating that an increase in NADPH oxidase activity mediates these effects. The slight reduction in O₂^{• −} formation by allopurinol suggests that xanthine oxidase may also be upregulated.

These data are in agreement with previous observations. IL–1β has been shown to induce O₂^{• −} production from cultured rat pulmonary microvascular smooth muscle cells, human internal mammary artery smooth muscle cells, and human cord vein endothelial cells.^15,16 TNF-α activated O₂^{• −} producing NADH oxidase in cultured rat aortic smooth muscle cells in a time and dose dependent manner.¹⁷ Increased O₂^{• −} has also been reported in whole lungs from guinea pigs following the injection of TNF-α.¹⁸ LPS challenge also increases lung O₂^{• −} generation in anaesthetised rats, which was found to coincide with the alterations in cardiovascular function measured in these animals.¹⁹ Clinically, it has long been recognised that intrapulmonary oxidative stress is present in patients with ARDS.^20,21

Removal of the endothelium resulted in a marked increase in O₂^{• −} release from PA segments both before and after incubation with LPS, TNF-α, and IL-1α, indicating that the endothelium of PAs possesses a system for removing O₂^{• −} radicals. Although there are reports that the endothelium produces O₂^{• −},²² our results show that the endothelium quenches the output of O₂^{• −} by the arterial segment as a whole. Since the endothelium is a principal source of NO, the reaction of O₂^{• −} with NO may account for the removal of O₂^{• −} by the endothelium and, as such, was investigated using NO synthase (NOS) inhibitors. The non-specific NOS inhibitor l-NAME mimicked the effect of endothelium removal—that is, O₂^{• −} release was increased by l-NAME in segments with intact endothelium—indicating that NO removes O₂^{• −} from the system through the reaction: NO + O₂^{• −} → ONOO⁻, thereby reducing that which reacts with ferricytochrome c. Indeed, immunocytochemical analysis clearly showed a marked increase in NT (an index of ONOO⁻ formation) in the endothelial region of PA segments after incubation with cytokines and LPS, supporting this proposal.

To differentiate between the roles of eNOS and iNOS in reducing O₂^{• −} formation in intact PAs, the effect of two inhibitors of these isoforms were investigated. The eNOS inhibitor l-NIO, but not the iNOS inhibitor l-NIL, enhanced O₂^{• −} formation in response to IL-1α and TNF-α and, to a lesser extent, to LPS. Immunocytochemical analysis confirmed that eNOS (but not iNOS) is upregulated in the endothelium of the arterial segments in response to these cytokines and that LPS had no effect. It therefore appears that cytokines (principally) elicit a simultaneous upregulation of both smooth muscle NADPH oxidase and endothelial eNOS which negates the increased formation of O₂^{• −}, thereby protecting the vessel against oxidative damage. In turn, loss of the endothelium would render the PA susceptible to the pathogenic impact of cytokine induced O₂^{• −} formation. These data are also in agreement with those of Bhagat et al²³ who concluded that IL-1β upregulates eNOS but not iNOS in venous tissue of human subjects in vivo.

In contrast, the iNOS inhibitor l-NIL enhanced O₂^{• −} formation in response to LPS but had no effect on IL-1α or TNF-α mediated O₂^{• −} formation. Immunocytochemical analysis again confirmed an upregulation of iNOS in response to LPS but no effect of cytokines on this isoform. It therefore appears that LPS has little effect on the upregulation of eNOS but influences the expression of iNOS at the level of the vascular smooth muscle. Nevertheless, NO produced by iNOS appears also to have a protective effect against LPS induced O₂^{• −} formation. Other studies have shown that LPS promotes the upregulation of iNOS while suppressing the expression of eNOS.^24,25 Additionally, as these data indicate, the excess formation of ONOO⁻ may contribute to the pathophysiology of ARDS directly since ONOO⁻ has been shown to exert pathological effects including endothelial cell apoptosis and the expression of adhesion molecules.^26,27 Clinically, raised levels of NO metabolites are present in high levels in bronchoalveolar lavage (BAL) fluid from patients with ARDS, which is consistent with an upregulation of eNOS and iNOS in intrapulmonary tissues in this condition.²⁸ More crucially, increased levels of nitrotyrosine residues (index of ONOO⁻) have been reported in BAL fluid from patients with acute lung injury,^29,30 consolidating the hypothesis that upregulation of both NO and O₂^{• −} generating enzymes occurs in ARDS by the mechanisms proposed in this study.

In conclusion, the present study shows that LPS and two major cytokines additively promote the formation of O₂^{• −} by pig pulmonary arteries in vitro, principally through an upregulation of NADPH oxidase. In turn, this O₂^{• −} would perpetuate and augment inflammation through the induction of adhesion molecule expression, vasoconstriction, and the negation of NO bioavailability. In this situation the endothelium reduces LPS and cytokine induced O₂^{• −} generation, a mechanism that appears to be mediated through NO derived from eNOS (in the case of cytokines) and iNOS (in the case of LPS). The concomitant expression of these NOS isoforms with that of NADPH oxidase may therefore constitute an important protection system against oxidative stress. However, at some point in the sequelae of ARDS this defence system may be overwhelmed, rendering the vasculature susceptible to oxidative stress. The integrity of the endothelium may thus be axiomatic in the progression and severity of ARDS. A reduction in oxidative stress coupled with the delivery of NO would therefore seem to be a potentially effective strategy for treating ARDS.