We employed a bile duct ligation (BDL) model of cholestatic liver injury to test the hypothesis that this form of preexisting hepatic dysfunction alters the kinetics of circulating TNF-alpha and IL-6 after Escherichia coli endotoxemia, thereby augmenting mortality and lung injury by a TNF-alpha:leukotriene (LT) axis of inflammation. Male rats were catheterized 13 d after BDL or sham surgery and studied while awake 18 to 24 h later. Cholestasis after BDL was confirmed by baseline serum bilirubin (BDL = 7.34 +/- 0.72 mg/dl, mean +/- SEM, n = 17 versus Sham = 0.25 +/- 0.07, n = 20; p < 0.005) and histopathology. Sham and BDL animals received E. coli lipopolysaccharide serotype O55:B5 (LPS, 5 mg/kg i.v.) or 0.9% NaCl (NS) ending at t = 0 and were monitored over 24 h for vital signs and hemodynamics. In parallel studies, lipoxygenase inhibition was performed using diethylcarbamazine or the 5-lipoxygenase activating-protein inhibitor MK-886. Blood was collected at baseline and at t = 1.5, 3.5, and 24 h for formed elements and for serum endotoxin, TNF-alpha, IL-6, bilirubin, and alanine aminotransferase (ALT). Organs were evaluated at 24 h for histopathology, including neutrophil (PMN) densities and wet/dry weight (W/D) ratios. Cholestasis reduced survival after otherwise nonlethal endotoxemia, with seven of 11 BDL + LPS rats dying within 24 h versus no deaths in BDL + NS (n = 6), Sham + LPS (n = 14), or Sham + NS (n = 6) animals (p < 0.01). Despite equivalent serum endotoxin between groups, circulating TNF-alpha was 8-fold higher in BDL + LPS than in Sham + LPS rats at 1.5 and 3.5 h (p < 0.001), whereas serum TNF-alpha did not differ between BDL + NS and Sham + NS rats. IL-6 likewise was increased differentially by 1.5 h in BDL + LPS animals (11.98 +/- 2.42 ng/ml) versus Sham + LPS rats (3.05 +/- 0.58 ng/ml, p < 0.05). Hypothermia, bradycardic hypotension, and leukopenia were most severe and prolonged in BDL + LPS rats, which also had significantly higher ALT values, W/D ratios, and organ PMN counts. LT inhibition failed to reduce BDL-related differences in serum cytokines or survival after endotoxemia. Thus, cholestasis augments inflammatory responses to gram-negative endotoxemia, sensitizing the host to enhanced fluid flux in multiple organs and to mortality by a LT-independent mechanism.

Hepatic cytokine gene expression is independently stimulated by circulating microbial products and reductions in the cellular O2 supply. Although these stimuli occur sequentially after gram-negative bacteremia, it is unknown whether their interplay augments production of interleukin (IL)-1 by the liver. We studied the effects of intraportal Escherichia coli (EC) bacteremia and secondary constant-flow hypoxia (Po2, approximately 46 Torr for 30 min) on IL-1 alpha and IL-1 beta gene expression in ex situ buffer-perfused rat livers over 180 min (n = 67). At t = 0, normoxic EC and normal saline (NS) controls received 10(9) live EC serotype 055:B5 and 0.9% NaCl, respectively; in livers subjected to EC+hypoxia-reoxygenation (H/R) and NS+H/R, hypoxia began 0.5 h after EC or NS and was followed by 120 min of reoxygenation. Portal and hepatic venous perfusates were serially analyzed for bacterial colony-forming units, O2 uptake, and aspartate aminotransferase. At 60 min (peak hypoxia) and 180 min, cDNAs for IL-1 alpha and IL-1 beta were hybridized to whole liver RNA, and IL-1 beta protein levels in venous perfusates were assessed. Intrahepatic levels of reduced glutathione (GSH) were measured as an index of oxidative stress. Compared with normoxic EC, IL-1 alpha transcripts decreased at 180 min in EC+H/R livers (P < 0.0001) as did IL-1 beta mRNA (P < 0.05), despite similar EC clearance, GSH levels, posthypoxic O2 uptake, and aspartate aminotransferase release. Hepatic secretion of IL-1 beta likewise fell in EC+H/R vs. EC controls (P < 0.005). Prostaglandin H synthase-2 (COX-2) message accumulation was not enhanced by H/R, and indomethacin did not reverse H/R-mediated suppression of IL-1 production. In contrast, H/R-related falls in EC-induced IL-1S expression were reversed by allopurinol or catalase. Thus brief hypoxic stress of the liver causing neither GSH depletion nor functional impairment downregulates postbacteremic IL-1 expression by a mechanism involving O2 radicals but not cyclooxygenase metabolites.

We hypothesized that ischemia/reperfusion (I/R) injury of the liver during normotensive gram-negative bacteremic sepsis alters the kinetics of circulating endotoxin, tumor necrosis factor-alpha (TNF-alpha), and coinduced mediators, thereby exacerbating sepsis-induced lung inflammation. Liver and lung dysfunction were studied after hematogenous infection of Sprague-Dawley rats with 10(9) Escherichia coli serotype O55:B5 (EC) and 90 min of secondary hepatic ischemia in EC + I/R and saline-infused (normal saline NS) x I/R rats, followed by brief (1 h) or longer reperfusion (24 h). TNF- alpha:leukotriene interactions in this model were examined using the 5-lipoxygenase-activating protein inhibitor MK-886. Compared with sham-operated EC + Sham animals, peak serum endotoxin, TNF-alpha, alanine aminotransferase, interleukin-6 (IL-6), and hepatic neutrophil (PMN) influx were higher in EC + I/R rats through 24 h (p < 0.05) despite comparable arterial pressure. Lung PMN influx and wet/dry weight ratios were likewise enhanced in EC + I/R versus EC + Sham or NS + I/R rats. MK-886 attenuated TNF-alpha concentrations and ischemic liver injury but not mortality. Thus, focal hepatic I/R augments circulating endotoxin, TNF-alpha, and postbacteremic lung inflammation early after normotensive E. coli bacteremic sepsis.

Reductions in hepatic O(2) delivery are common early after gram-negative bacteremic sepsis owing to cardiopulmonary dysfunction and derangements in sinusoidal perfusion. Although gram-negative endotoxin and cellular hypoxia independently enhance activation of nuclear factor-kappaB (NF-kappaB) via generation of reactive O(2) species (ROS), the combination of these stimuli downregulates hepatic TNF-alpha gene expression. Here we tested the hypothesis that hypoxic suppression of postbacteremic TNF-alpha gene expression is transcriptionally mediated by reduced activation of NF-kappaB. Buffer-perfused rat livers (n = 52) were studied over 180 min after intraportal infection at t = 0 with 10(9) live Escherichia coli (EC), serotype O55:B5, or 0.9% NaCl controls under normoxic conditions, compared with 0.5 h of constant-flow hypoxia (PO(2) approximately 41 +/- 7 Torr) beginning at t = 30 min, followed by 120 min of reoxygenation. In parallel studies, tissue was obtained at peak hypoxia (t = 60 min). To determine the role of xanthine oxidase (XO)-induced ROS in modulating NF-kappaB activity after hypoxia/reoxygenation (H/R), livers were pretreated with the XO inhibitor allopurinol, with results confirmed in organs of tungstate-fed animals. Electrophoretic mobility shift assays were performed on nuclear extracts of whole liver lysates using (32)P-labeled oligonucleotides specific for NF-kappaB. Compared with normoxic EC controls, hypoxia reduced postbacteremic NF-kappaB nuclear translocation and TNF-alpha bioactivity, independent of reoxygenation, tissue levels of reduced glutathione, or posthypoxic O(2) consumption. XO inhibition reversed the hypoxic suppression of NF-kappaB nuclear translocation and ameliorated decreases in cell-associated TNF-alpha. Thus decreases in hepatic O(2) delivery reduce postbacteremic nuclear translocation of NF-kappaB and hepatic TNF-alpha biosynthesis by signaling mechanisms involving low-level generation of XO-mediated ROS.