INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Oxygenation in acute lung injury (ALI) is strongly affected by the postinjury ventilation/perfusion (/) pattern. In at least some experimental models of ALI (1, 2), pulmonary blood flow (PBF) redistributes away from the injured edematous lung regions ("perfusion redistribution"), thereby preserving a relatively normal / pattern and minimizing the decrease in Pa_O2 (3, 4). In contrast, when perfusion redistribution fails to develop after ALI, the deterioration in oxygenation is much worse (3, 4). The mechanisms responsible for any given perfusion pattern in ALI have not been clearly determined. Previous experimental studies, however, suggest that the vasoactive effects of eicosanoids are likely to be involved in ALI (1).

Both thromboxane and prostacyclin are produced by prostaglandin G/H synthase, also known as cyclooxygenase (COX). Two isoforms of COX have been described: COX-1, which is constitutively expressed in most tissues, and COX-2, which can be induced in certain cell types by inflammatory stimuli (10, 11). In a previous study we found that ALI was associated with increased tissue concentrations of the stable thromboxane metabolite TXB₂, but that circulating levels of TXB₂ were not consistently affected (1). Thromboxane receptor blockade is associated with reduced formation of pulmonary edema, but the pulmonary perfusion pattern is unaffected (8). The source of the increased tissue thromboxane in lung tissue could be either release from platelets or increased synthesis from cells resident in the lung. Changes in circulating eicosanoid concentrations, however, are especially striking when ALI is preceded by a priming (low) dose of endotoxin (Etx) (3). Endotoxin, in turn, is known to upregulate the expression of COX-2 (12). Increased levels of circulating prostacyclin due to Etx priming have been reported to eliminate perfusion redistribution in this model of lung injury (3, 4). Both the changes in tissue thromboxane levels and in circulating prostacyclin in ALI, along with the physiologic effects on edema formation and pulmonary perfusion pattern, could be due to an increased COX-2-mediated synthesis of thromboxane in lung tissue. The recent development of highly specific COX-2 inhibitors made it possible for us to test this hypothesis (13).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Preparation

Our study was approved by the Animal Studies Committee of the Washington University School of Medicine. Twenty-one adult mongrel dogs (weight: 18 to 23 kg) were anesthetized with pentobarbital sodium (40 mg/kg), intubated with a cuffed endotracheal tube, and ventilated (FI_O2 = 1.0) with a Harvard pump respirator (Harvard Co., South Natick, MA) at a tidal volume (VT) of 15 ml/kg, with the respiratory rate adjusted to achieve a normal arterial Pa_CO2. Positive end-expiratory pressure was not used. Additional barbiturate was given, as necessary to eliminate spontaneous breathing.

Instrumentation was performed with animals in the supine position. After percutaneous insertion of two size 8.5 French introducers (Baxter, Irvine, CA) in both femoral veins, a size 7.5 French balloon-tipped pulmonary artery catheter (Baxter) and a 110-cm size 7 French pigtailed catheter (Cook, Bloomington, IN) were positioned in the pulmonary artery under fluoroscopic visualization. A 20-gauge artery catheter (Arrow, Reading, PA) was placed into the right femoral artery via the Seldinger technique for blood sampling; a size 6.0 French introducer (Cook) was percutaneously inserted into the right external jugular vein for drug and radionuclide administration. Catheter patency was maintained by periodic infusion of heparinized saline (1 U/ml).

Positron Emission Tomography Techniques

Measurements made with positron-emission tomography (PET) were obtained with an in-house-built "Super-PETT" 3000 system. The design features, methods for calibration, corrections for activity decay, and corrections for photon attenuation of the system have been described previously (16).

The animals were placed in the PET scanner, with the most caudal tomographic slice being located about 1 to 2 cm below the level of the dome of the diaphragm. Data were recorded simultaneously from seven slices with a center-to-center separation of 1.05 cm and an in-plane, full-width, half-maximum spatial resolution of 0.85 cm. The image reconstruction resolution was set at 12 mm.

The methods used to measure pulmonary blood flow, including supporting validation studies, have been described previously in detail (16, 20). In general, PET is used to measure the tissue concentration and distribution of a positron-emitting radionuclide, which in the present study was simply H₂¹⁵O. The activity data measured with PET, when combined with blood activity (used as a reference) and analyzed with an appropriate compartmental mathematical model, yield tomographic images representative of PBF.

Imaging Protocols

Because the pattern of PBF in normal dogs (1, 2, 4, 7, 8, 21) is well known, and because we recently reported twice in detail the pattern of PBF in oleic acid (OA)-induced ALI after low-dose Etx priming (3, 4), regional PBF was measured only with positron emission tomographic imaging in groups receiving OA only, the novel COX-2 inhibitor SC-236 + OA, and SC-236 + Etx + OA (Figure 1). In the OA-only group (n = 6), the only intervention was the administration of 0.08 ml/kg OA (Fisher Scientific, Pittsburgh, PA), given into the central venous catheter. In the SC-236 + OA group (n = 3), the SC-236 was given intravenously in a dose of 10 mg/kg at 50 min before central venous injection of OA. The drug was a generous gift from G. D. Searle and Co. (Skokie, IL). In vitro, SC-236 is a highly specific inhibitor of human recombinant COX-2 as compared with COX-1 (IC₅₀ for COX-1: 17.8 µM; IC₅₀ for COX-2: 0.01 µM) (15). In the SC-236 + Etx + OA group (n = 5), 10 mg/kg of SC-236 was given intravenously 20 min before the administration of 15 µg/kg of Escherichia coli Etx (Fisher Scientific) by injection, which was in turn followed by an OA injection after an additional 30 min. This dose of Etx is known to eliminate hypoxic pulmonary vasoconstriction and perfusion redistribution during ALI without causing significant systemic hemodynamic changes (3, 4).

View larger version (10K): [in this window] [in a new window]   Figure 1.   Experimental timeline. Asterisks indicate timing of blood sample collections for eicosanoid analysis. In the four experimental groups, baseline blood was drawn just before administration of either the COX-2 inhibitor SC-236 or placebo (P) (first upright arrow = SC-236/P). Second upright arrow (E/P) indicates timing by which either endotoxin (E) or placebo was given. Third upright arrow (OA/P) indicates timing by which either OA or P was administered.

Blood samples for eicosanoid analysis (see the subsequent discussion) were obtained after instrumentation was completed ("baseline"), before a background ("blank") PET scan (Figure 1). The following scans were then done: (1) a transmission scan used to correct for photon attenuation during emission scans, and for the placement of regions-of-interest (ROIs) for later image analysis (see the subsequent discussion); (2) a 15- to 18-s scan (used for PBF measurement) obtained during a continuous infusion of about 60 mCi of H₂¹⁵O (¹⁵O-labeled water; and (3) a 300-s scan obtained after equilibration of the H₂¹⁵O (for measurement of the apparent blood-tissue partition coefficient for water used in the PBF calculation).

After the first PET scan, either the COX-2 inhibitor or its vehicle was administered (Figure 1). After a second blood sample was withdrawn for eicosanoid analysis 20 min later, either Etx or its vehicle was given. After a third blood sample was withdrawn for eicosanoid analysis 30 min later, either OA or its vehicle was administered. After another 120 min, a second set of PET scans was performed and a fourth blood sample was withdrawn for eicosanoid analysis. After this, two samples of lung tissue were collected from each dog to evaluate lung tissue production of thromboxane and prostacyclin. Each dog was then euthanized with additional pentobarbital followed by 20 ml of saturated KCl.

Blood Biochemistry

Blood samples were analyzed with an enzyme immunoassay technique (Figure 1) for the stable metabolites of TXB₂ and prostacyclin (6-keto prostaglandin F₁ [PGF₁]) in the three imaging protocol groups, and in another group (n = 4) given endotoxin alone before OA (i.e., no SC-236). At each time point, a sample was obtained by drawing blood into a tube with ethylenediamine tetraacetic acid (EDTA) (1 mg/ml) and indomethacin (5 µg/ml). Blood samples were centrifuged at 5° C at 2,200 × g for 10 min. The plasma was removed and stored frozen at 30° C until assay.

Enzyme immunoassay of 6-keto PGF₁ and TXB₂ was done in 96-well microtiter plates precoated with goat anti-rabbit immunoglobu-lin G at 2 µg/well (22). Before use, the plates were washed with 10² M phosphate buffer (pH 7.4) containing 0.05% Tween 20 (wash buffer). The assay was performed in a total volume of 150 µl. In brief, 50 µl of acetylcholinesterase-conjugated eicosanoid tracer (Caymen, Ann Arbor, MI), 50 µl of antiserum directed against 6-keto PGF₁ or TXB₂ (PerSeptive Biosystems, Framingham, MA), and 50 µl of a standard or sample in assay buffer were combined and incubated at 25° C for 18 to 20 h. After the plates were washed three times with wash buffer, Ellman's reagent (200 µl) was dispensed into each well. Absorbance was recorded at 412 nm in a microtiter plate spectrophotometer (BioTech, Winooski, VT) when the absorbance for the well containing the zero standard (B₀) exceeded 0.200 absorbance units. Each sample was assayed in duplicate. A standard curve was generated for each assay. Sample eicosanoid concentrations were determined by comparison with a log-logit transformation of the standard curve. Eicosanoid concentrations were expressed as pg/ml blood.

Lung Tissue Eicosanoids

Lung tissue samples weighing 2 to 3 g were collected from dorsal and ventral regions of the lower lobes of both lungs. Immediately after excision, the tissue samples were placed in 40 ml of ice-cold methanol and homogenized with a Polytron homogenizer (Brinkmann Instruments, Inc., Westbury, NY). The tissue homogenate was then cooled to 30° C. Precipitated proteins were removed from the homogenate by centrifugation at 1,800 g for 30 min at 15° C. The clear methanolic extract was evaporated to dryness under vacuum at 30° C in a Speed-Vac concentrator (Savant Instruments, Inc., Holbrook, NY). Samples for eicosanoid assay were solubilized in phosphate buffer (0.1 M, pH 7.4), containing 0.4 M NaCl, 10³ M EDTA, 0.1% bovine serum albumin, and 0.01% sodium azide (assay buffer). Eicosanoid concentrations were measured as described earlier and expressed as ng/g lung tissue.

Image Analysis

From each dog (except those in the E + OA group), the four contiguous tomographic slices with the greatest lung volume were analyzed from the seven slices reconstructed as part of each PET scan, encompassing most of the caudal lobes. ROIs from the right and left lungs were defined on each transmission scan, as previously described (1, 8).

The position of each region was kept in computer memory, and mean values for each region were obtained for all PET measurements made. PBF was measured as ml/min/100 ml lung. To normalize the regional PBF data for differences in cardiac output, PBF in each picture element (pixel) was expressed as a fraction of the total blood flow to the region.

To evaluate the relationship of PBF to anatomic position within a region, the x and y coordinates for each pixel, along with the respective fractional PBF values for each pixel, were recorded. The pixel data were then sorted, first by their y-coordinate values. Next, within each value for y, the data were sorted again by their x-coordinate values. The result was a listing of the pixels by location, beginning in the most ventromedial portion of the region and ending with the most dorsolateral portion of the region. Each region contained ~ 400 to 500 pixels. Arbitrarily, the data were divided into 20 "bins" stacked vertically in the ventral-dorsal direction, so that each bin contained ~ 20 to 25 pixels, which could then be averaged. By keeping the number of bins per region and the number of tomographic slices per dog constant, bin values could be averaged across dogs, allowing comparisons between experimental groups.

To quantify perfusion redistribution, we determined the difference in fractional PBF between the baseline and subsequent PET data sets, and summed the difference in those bins in which PBF decreased between the two time points (1, 8).

Statistical Analysis

Data are presented as means ± SD. For PET and plasma eicosanoid data, statistical significance was determined with a repeated measures analysis of variance (ANOVA) using the general linear models procedure of the Statistical Analysis System (SAS Institute, Cary, NC). Post hoc testing with Tukey's honestly significant difference (HSD) test was limited to comparisons of baseline data, to changes from baseline data among the four experimental groups, and to differences among the groups at the same time point. The thromboxane and prostacyclin synthesis rate data were analyzed through the one-way ANOVA procedure of the Statistical Analysis System. Post hoc testing was performed with Duncan's multiple range test. We accepted a value of p < 0.05 as indicating statistical significance.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effect of COX-2 Inhibition on Lung Perfusion Pattern

The baseline perfusion pattern was similar in all three groups in which PET imaging was obtained (Figure 2A). After administration of OA, perfusion to edematous dorsal lung regions decreased, regardless of pretreatment with either SC-236 or SC-236 + Etx (Figure 2B, Table 1). Thus, COX-2 inhibition had no significant effect on the expected changes in perfusion pattern caused by OA-induced lung injury alone. However, COX-2 inhibition did block the expected vasodilatory response to Etx, as reported in two recent studies (3, 4). In other words, in this model, pretreatment with low-dose Etx is expected to block perfusion redistribution as measured by PET imaging (3, 4). However, this effect of low-dose Etx was itself inhibited by prior treatment with the COX-2 inhibitor SC-236, allowing perfusion redistribution to occur (Figure 2, Table 1).

View larger version (15K): [in this window] [in a new window]   Figure 2.   Average ventral-dorsal distribution of fractional PBF in three experimental groups studied with PET imaging: OA only, SC-236 + OA, and SC-236 + Etx (E) + OA. Bin numbers represent equal collections of picture elements on multiple PET images of PBF, with lower bin numbers situated in ventral regions and higher bin numbers situated in the dorsal regions of the lungs. Each symbol represents the mean value for all dogs in each group. (A) Baseline. (B) At 120 min after OA-induced lung injury. At either time there is no significant difference in the pulmonary perfusion pattern among the groups, although there is a significant reduction of relative PBF (perfusion redistribution) in the dorsal lung regions (the injured regions that develop the greatest increase in lung water) after administration of OA despite pretreatment with either the COX-2 inhibitor SC-236 or with SC-236 + Etx (E). For reference, the baseline distribution pattern of the OA only group is shown as a solid line in B.

Effect of COX-2 Inhibition on Plasma Eicosanoid Levels

At baseline, plasma TXB₂ levels were comparable in all groups. Intervention with SC-236, Etx, and/or OA infusions had no consistent effect on circulating TXB₂ levels (Figure 3A).

View larger version (17K): [in this window] [in a new window]   Figure 3.   (A) Plasma TXB2 (stable metabolite of thromboxane) and (B) 6-keto PGF1 (stable metabolite of prostacyclin [PGI2]) in four experimental groups. "Baseline" data were obtained before giving the COX-2 inhibitor SC-236 or placebo (P), whereas "After SC-236/P" data were obtained just before administration of either Etx (E) or placebo, and "After E/P" data just prior to giving OA. "After OA injury" data were obtained 120 min after OA administration. There is no significant difference in circulating TXB2 levels among the groups at any time. However, after administration of both low-dose Etx and OA, 6-keto PGF1 plasma levels increased statistically significantly by more than 17 times over the baseline value, and by nearly four times the levels seen in the other three experimental groups (asterisk). Because of inhibition of COX-2, plasma levels of 6-keto PGF1 did not change significantly in the group "SC-236 + E + OA" either after Etx or after OA administration.

The administration of SC-236 before either Etx or OA had no significant effect on circulating levels of 6-keto PGF₁ (Figure 3B). Etx alone (in the absence of SC-236) had only a small effect on circulating PGF₁ levels (Figure 3B). However, as previously reported (3), the combination of low-dose Etx with subsequent OA-induced lung injury resulted in a marked effect on circulating levels of 6-keto PGF₁, reaching more than 17 times the baseline value (Figure 3B). This synergistic effect between Etx and lung injury was completely blocked by pretreatment with SC-236 (Figure 3B).

Effect of COX-2 Inhibition on Lung Tissue Eicosanoid Concentrations

As previously reported (1), OA-induced injury results in significantly (p < 0.05) increased lung tissue TXB₂ concentrations as compared with those in uninjured lung (Figure 4A). This effect was not altered by SC-236 (Figure 4A).

View larger version (17K): [in this window] [in a new window]   Figure 4.   Concentrations of TXB2 (A) and prostacyclin (6-keto PGF1) (B) in lung tissue after OA-induced lung injury. Data from the current study (filled bars) are compared with one another and with data from a previous study (empty bars) (1). (A) The levels of TXB2 after OA are increased in all cases as compared with those in uninjured lung tissue. The levels are highest after Etx pretreatment (E + OA group). The COX-2- specific inhibitor SC-236 has no significant effect on the lung tissue TXB2 levels (p < 0.05 for the E + OA group as compared with the "OA only," "SC-236 + OA," and "no injury" groups, but not with the "SC-236 + E + OA" group). (B) The levels of 6-keto PGF1 in lung tissue were increased only when OA-induced injury followed pretreatment with low-dose Etx (E + OA group). This effect was completely blocked by SC-236 (*p < 0.05 for the E + OA group compared with each of the other groups).

Pretreatment with low-dose Etx caused an even greater increase in lung tissue TXB₂ concentrations (p < 0.05) (Figure 4A). This effect was also not significantly affected by pretreatment with SC-236.

As previously reported (1), OA-induced injury did not result in any significant change in lung tissue 6-keto PGF₁ concentrations (Figure 4B). Pretreatment with SC-236 did not alter this finding. However, pretreatment with low-dose Etx resulted in a highly dramatic increase in the lung tissue concentrations of 6-keto PGF₁ (p < 0.05). In contrast to the situation with TXB₂ concentrations in OA-induced lung injury, this effect was completely blocked by pretreatment with SC-236 (Figure 4B) (p < 0.05).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The major findings of this study were that selective COX-2 inhibition: (1) maintains perfusion redistribution during ALI despite endotoxin priming; and (2) prevents the increase in circulating and lung tissue prostacyclin that is the result of endotoxin priming in ALI.

Even though thromboxane does not seem to be an important determinant of the pulmonary perfusion pattern in the OA-induced lung injury model (2), thromboxane receptor blockade does reduce edema-fluid accumulation in the lung, suggesting that thromboxane has a potentially important role in the pathogenesis of pulmonary edema (2). We previously reported that lung tissue thromboxane concentrations were increased by lung injury and that we could block this effect with the nonspecific COX inhibitor meclofenamate (1). This finding could have been due to: (1) increased substrate (i.e., arachidonic acid) availability resulting from altered phospholipase activation; (2) a change in the amount and/or activity of COX in resident lung cells; (3) an increased release of thromboxane by platelets in lung tissue; and/or (4) tissue thromboxane accumulation (i.e., "trapping") in alveolar edema fluid during the period of experimental observation prior to the obtaining of tissue.

In the present study, and in contrast to the work just described with nonspecific COX inhibitors (1, 3), the specific COX-2 inhibitor SC-236 did not decrease the accumulation of TXB₂ in lung tissue (Figure 4). Since platelets are a rich source of thromboxane but contain only COX-1 (12, 25), pretreatment with a specific COX-2 inhibitor would not be expected to affect any putative platelet effects mediated by increased thromboxane production. However, our study cannot rule out other resident lung cells like monocytes, vascular endothelial cells, smooth muscle cells, mast cells and/or alveolar macrophages as alternative or additional source(s) of the increased thromboxane in lung tissue during ALI.

A priming (low) dose of Etx caused a moderate (~ 4-fold) increase in lung tissue TXB₂ levels (Figure 4). Since this effect also was not significantly altered by pretreatment with SC-236, the increase in TXB₂ in injured lung is probably not due to increased COX-2 expression. Since Etx is not known to alter COX-1 regulation, the observed increase in TXB₂ may represent increased platelet trapping in the lung in response to OA-induced injury after pretreatment with Etx. Such an explanation is consistent with the observation that increased TXB₂ levels were not detected in plasma in response to Etx (Figure 3).

As previously reported, a priming (low) dose of Etx has dramatic effects on the physiology of the OA-induced lung injury model (3, 4)effects that cannot be attributed to Etx alone but are due to a synergistic effect between Etx and lung injury (3). One prominent effect is that the expected redistribution of pulmonary perfusion away from edematous lung regions is eliminated. This effect is associated with increased levels of prostacyclin in plasma (Figure 3) and lung tissue (Figure 4). The effects of low-dose Etx on both eicosanoid production and on the distribution of PBF can be prevented with the nonspecific COX inhibitor meclofenamate (3). The results of the current study, with the selective COX-2 inhibitor SC-236, included both of these effects, suggesting that they are specifically mediated by the inducible form of COX.

These results are consistent with findings by other investigators. For instance, Ogletree and coworkers reported that COX inhibitors affect pulmonary vascular responses to Etx in sheep (29). Spannhake and associates reported similar results, and found that pulmonary COX activity increased in dogs after Etx administration (30). These earlier studies, however, were intended as models of septic shock or Etx-induced ALI. The dose of Etx used in the current study is not associated with these hemodynamic or pulmonary effects when administered in isolation (3).

Our earlier studies preceded the more recent recognition that COX can be expressed as different isoforms. Ermert and coworkers recently reported that COX-2 as well as COX-1 is expressed in several cell types of normal rat lung, and that COX-2 may also be involved in the regulation of vasomotor tone in the normal lung (31, 32). Furthermore, previous in vitro studies provide convincing evidence that an increase in COX-2 gene and protein expression can be induced with interleukin-1, tumor necrosis factor-, or lipopolysaccharide (Etx) (33), agents associated with other models of lung injury. In addition, acute hypoxia (34) and oxidant stress (35), both present in many models of lung injury, have been associated with an increase in COX-2 activity. These results suggest that an increase in COX activity in either OA-induced ALI alone or after a priming dose of Etx could be caused by the induction of COX-2. The data gathered in the present study, done with a highly selective COX-2 inhibitor, provide some of the first direct evidence that COX-2 can indeed be involved in the physiologic changes accompanying ALI, especially when ALI is associated with (even minor degrees of) endotoxemia.

In summary, the current study demonstrates that COX-2 mediates the effect of a priming dose of Etx on prostacyclin production and pulmonary perfusion, but not lung tissue thromboxane accumulation, after OA-induced ALI.