INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

MULTIPLE ORGAN DYSFUNCTION syndrome (MODS) is the leading cause of death in critically ill patients with gram-negative bacteremic sepsis (37). Recently, sepsis-related organ damage culminating in MODS has been postulated to result from overexpression of multiple inflammatory cytokines in response to gram-negative endotoxin. Of these, tumor necrosis factor- (TNF-) and interleukin-1 (IL-1) have been implicated as proximal mediators of sepsis-induced lung injury and MODS (6, 24). The liver, containing the largest resident macrophage population in the body, is increasingly recognized as a key regulatory organ, producing and exporting such inflammatory cytokines in response to bacterial products or reductions in blood flow or oxygen delivery (8, 11).

Critical to the expression of TNF- and TNF--induced secondary mediators of inflammation is the NF-B/c-Rel family of nuclear transcription factors, which are activated during sepsis (5). NF-B, a heterodimer consisting of p65 and p50 subunits, is bound to IB, its cytosolic inhibitor protein. After a wide array of stimuli, including bacterial products and changes in the cellular O₂ supply, IB is phosphorylated and degraded, thereby promoting the nuclear translocation of NF-B. By binding to specific cis-acting elements in the promoter regions of TNF- and IL-1, cytokine-specific gene expression is enhanced (2). Inasmuch as most studies of this process have been performed in cultured cells, the mechanisms responsible for the activation and nuclear translocation of NF-B after postbacteremic oxidative stress are not well understood at the intact organ level. Even so, the generation of reactive oxygen species (ROS) in excess of intracellular antioxidant defense systems, leading to increased tyrosine kinase activation and IB phosphorylation, is thought to be the final common pathway regardless of the initial stimulus or combination of stimuli (2, 9, 31).

In this context, the critical role of ROS in signal transduction and NF-B activation is recognized (25, 31), although the necessary intracellular concentrations of ROS and their interplay with endogenous antioxidants are incompletely characterized. Under normoxic conditions, gram-negative endotoxin has been demonstrated to stimulate production of ROS in perfused rat liver (3) concomitant with the activation of NF-B and followed by the subsequent production of cytokines (10, 34, 35). In a variety of cultured cells, periods of hypoxia and reoxygenation (H/R) alone are sufficient to activate NF-B, most likely by generating ROS, which function as intracellular messengers (26, 28). In support, antioxidants added to endotoxin-stimulated cells under normoxic conditions or to cells cultured under hypoxic conditions without prior endotoxin exposure inhibit NF-B activation and subsequent cytokine production (24, 33).

Despite the occurrence of bacteremia and reductions in oxygen delivery during sepsis, there are no studies to determine the effects of these sequential stimuli on NF-B activation. According to the current paradigm, additive generation of ROS by each of these conditions would be predicted to increase NF-B activation until reduced glutathionine (GSH) levels fall below a critical threshold, after which NF-B becomes reversibly oxidized and unable to bind to DNA (8, 29). However, we previously demonstrated in perfused liver that TNF- and IL-1 gene expression are downregulated when brief (30 min) hypoxic stress follows intraportal Escherichia coli (EC) infection despite unchanged hepatic GSH levels (20, 34). We now demonstrate that such brief and otherwise well-tolerated hypoxia potently reduces the postbacteremic activation of NF-B in the intact liver and that under these conditions, blockade of ROS generated via the xanthine oxidase pathway paradoxically restores the transactivation of NF-B and subsequent TNF- production. These results suggest that hypoxic suppression of postbacteremic TNF- gene expression is transcriptionally mediated by corresponding reductions in NF-B activation and further identify a novel anti-inflammatory role of ROS under mildly prooxidant conditions early after sepsis.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Materials. Minimum essential medium, RPMI 1640, actinomycin D, sodium taurocholate, glucagon, and other chemicals were from Sigma Chemical (St. Louis, MO) unless otherwise noted. Allopurinol (Allo), USP grade, was a gift from Burroughs-Wellcome (Research Triangle Park, NC). Molybdenum-deficient, tungstate-enriched normal protein diet chow (20% casein purified high nitrogen, 36.13% cornstarch, 10% corn oil, 3.5% CIN-76 mineral mix, 30% sucrose, 0.3% methionine, 0.07% sodium tungstate 2 H₂O) was purchased from ICN Pharmaceuticals (Costa Mesa, CA).

Animals. Pathogen-free adult male Sprague-Dawley rats (Harlan, Indianapolis, IN) weighing 260-325 g were allowed free access to food and water before experiments. In the tungsten diet group, animals initially weighing 100-125 g were fed the molybdenum-deficient, tungstate-enriched normal protein chow for 2-3 wk until their weights were similar to animals of other experimental groups. Studies were performed according to National Institutes of Health guidelines and were approved by the Animal Care Committee of Saint Louis University.

EC cultures. EC serotype O55:B5 was obtained as strain 12014 from the American Type Culture Collection (Bethesda, MD), maintained in suspension cultures of trypticase soy broth, and grown up over 18-24 h in fresh cultures. Organisms were sedimented at 1,000 g for 10 min at 4°C, washed twice in sterile 0.9% NaCl (NS), and resuspended in NS to 1 × 10⁹ colony forming units (cfu) in 1 ml, representing a 25% lethal dose at 24 h in conscious rats (20). Freshly prepared inocula were kept at 4°C until use and were vortexed immediately before intraportal infusion. Inocula enumerated as 1 × 10⁹ were shown by quantitative streak-plate cultures to average 9.2 ± 0.8 × 10⁸ cfu in all experimental groups.

Liver harvesting for ex situ perfusion. Anesthesia was induced with pentobarbital sodium (50 mg/kg ip), after which aseptic organ harvesting was performed in <15 min as previously described (20, 34). Briefly, the common bile duct and the portal vein were cannulated with a dripping 14-gauge Teflon catheter through which the liver was perfused during harvest with preoxygenated, heparinized (10 IU/ml) Krebs-Henseleit-Ringer bicarbonate (KHRB) buffer at ~12 ml/min. The thoracic inferior vena cava was cannulated, the liver was flushed clear of blood, and ex situ perfusion was begun at a flow rate of 3.75 ml · min¹ · g¹ and maintained at that flow rate for the remainder of the experiment. Organs showing incomplete blanching were discarded.

Experimental protocol. Perfusions were performed in the recirculating mode in a temperature-controlled (37 ± 0.5°C) system (20, 34). Briefly, perfusates were equilibrated with 95% O₂-5% CO₂ gas, resulting in baseline arterial PO₂ tensions of 550-620 mmHg. Before circuit priming (125 ml), perfusate was filtered (0.22 µm) and cultured on nutrient agar (24 h, 37°C) to confirm sterility. Perfusate endotoxin levels were 10 pg/ml by the Limulus assay (QCL-1000, Whittaker Bioproducts, Walkersville, MD). Perfusates were freshly prepared by addition of 10 mM glucose, 5% (vol/vol) sterile canine serum for opsonization, 10⁵ U penicillin, and 100 mg streptomycin to 1 liter of KHRB (final osmolarity, 297 ± 2 mosM and pH 7.4). Sodium taurocholate (50 µm) was added to 125 ml of perfusate; additional 50-µm aliquots were given every 30 min to replace bile acids (20, 34). Livers were equilibrated for 30 min before experiments to ensure steady-state conditions of O₂ consumption (O₂) and portal venous pressure. Supplemental NaHCO₃ was added as needed to maintain pH at 7.36-7.44.

1

1

7

Nuclear extraction. Immediately after experimental runs, liver sections were frozen in liquid N₂ and stored at 70°C. Nuclear extracts were prepared using a modification of the protocol described by Essani et al. (9). Briefly, 0.4-0.6 g of frozen liver were homogenized (EMI, Clinton, CT) in 3 ml ice-cold buffer containing (in mM) 10 HEPES, pH 7.9; 1.5 MgCl₂, 10 KCl, 1 dithiothreitol (DTT), 1 phenylmethyl-sulfonyl fluoride (PMSF), and 5 -glycerophosphate, and 10 µg/ml each of the protease inhibitors pepstatin, aprotinin, and leupeptin. After a 10-min incubation on ice, homogenates were centrifuged for 10 min at 850 g at 4°C. Supernatants were aliquoted and stored at 70°C for protein studies. The pellet was resuspended in 1 ml of ice-cold buffer containing the above reagents with 0.1% Triton X-100 and reincubated for 10 min. The samples were then vortexed and recentrifuged as above, the supernatant was again decanted, and the nuclear pellet was resuspended in 100 µl of an ice-cold buffer containing 20 mM HEPES, pH 7.9; 25% glycerol (vol/vol), 0.42 M NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 1 mM PMSF, 10 mM -glycerophosphate, and 10 µg/ml each of pepstatin, aprotinin, leupeptin. After being incubated on ice for 30 min, samples were centrifuged at 20,000 g for 30 min at 4°C. The supernatant containing the nuclear proteins was aliquoted and stored at 70°C. To avoid denaturation due to repeated freeze-thaw cycles, each thawed aliquot was used only once.

Electrophoretic mobility shift assays. Nuclear protein concentrations were determined by the bicinchoninic acid assay (Pierce, Rockford, IL) using bovine serum albumin as the standard. Ten micrograms of nuclear protein were incubated in binding buffer (10 mM Tris, pH 7.5; 1 mM EDTA, 5 mM MgCl₂, 5% glycerol, 5% sucrose, 0.01% Nonidet P-40) for 10 min at 37^°C. Two micrograms of poly dI:dC (Pharmacia Biotech, Piscataway, NJ) were added to each sample before the addition of the oligonucleotide probe. Double-stranded NF-B consensus oligonucleotide (5'-AGTTGAGGGGACTTTCCCAGGC-3') was end-labeled with [³²P]ATP using T4 polynucleotide kinase (Promega, Madison, WI), and ~100,000 counts/min were added to the reaction mixture. For competition studies, 2 µl of unlabeled NF-B oligonucleotide or of a noncompetitive oligonucleotide (NF-B mutant: 5'-AGTTGAGGCGACTTTCCCAGGC-3') were added before the addition of the radiolabeled probe. For supershift analysis, a 100-fold excess of antibodies cross-reactive to rat p65 and/or p50 subunits of NF-B were added to the reaction mixture. All oligonucleotides were purchased from Promega and antibodies from Santa Cruz Biotechnology (Santa Cruz, CA). After the addition of 1 µl of 10× gel-loading buffer to the reaction mixture, samples were run through a 4% acrylamide-bis gel in 1× running buffer (0.025 M Tris, 0.2 M glycine) at 250 V for 3-4 h in a 4°C room. Gels were vacuum-dried and exposed to X-ray film (Hyperfilm MD; Amersham, Arlington Heights, IL) for 24-72 h at 70°C, after which autoradiographs were developed. Individual bands were quantitated densitometrically (Molecular Dynamics, Sunnyvale, CA).

Intrahepatic GSH measurements. Concentrations of GSH in 500-mg frozen liver samples were measured in duplicate by high-performance liquid chromatography separation (Varian, Sunnyvale, CA). Fluorometric detection of the glutathione-o-phthalaldehyde adduct (Hitachi F-1050 Fluorescence Spectrophotometer, Danbury, CT) (20, 25) with results expressed as nanomoles per gram.

TNF- analyses. Cell-associated TNF bioactivity was quantitated with mycoplasma-free, actinomycin D-treated murine L929 cells (34). Standardized sections of frozen liver were homogenized with three volumes (vol/wt) of ice-cold homogenization buffer (RPMI 1640 supplemented with 4% bovine serum albumin, 1 mM DTT, 0.005 mM PMSF, 0.02% NaN₃, and 0.25 U/ml DNAase). Three milliliters of homogenate were layered over 2 ml of 41% sucrose solution and centrifuged (95,000 g, 30 min). The interfacial band containing the cellular membrane fraction was collected and resuspended in 5 ml of homogenization buffer. Membranes were then washed and pelleted twice by centrifugation (95,000 g, 30 min). The final pellet was resuspended in 1 ml of homogenization buffer, placed on ice, and used immediately for determination of TNF- bioactivity in duplicate over a fourfold dilution range on the basis of spectrophotometric measurement of L929 cell cytotoxicity at 550 nm. No interference with recombinant TNF--mediated cytolysis was observed when Allo was added to assay media over a proportional concentration range in vitro.

Statistical analyses. Serial changes in within-group variables were analyzed by a repeated-measure analysis of variance, with post hoc pairwise comparisons performed when appropriate by a Newman-Keuls test (20). Between-group comparisons of time- and group-specific variables were performed using a Kruskal-Wallis analysis, with pairwise comparisons made when appropriate using a two-tailed paired t-test. Significance was accepted at P < 0.05. Data are presented as means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Attenuation of postbacteremic NF-B activation by secondary H/R. In performing electrophoretic mobility shift assays (EMSAs) specific for NF-B, we found marked differences in NF-B activation after intraportal infection with EC compared with NS infusion (Fig. 1A). By 60 min after EC, enhanced nuclear translocation of NF-B compared with t = 0 was evident, with signals remaining elevated at t = 180 min. In contrast, time-matched signals from NS-infused livers did not differ from t = 0. Of note, hypoxic stress under nonseptic conditions in NS + H livers showed activation of NF-B at peak hypoxia (t = 60 min), which equalled that seen in time-matched EC-infected livers and did not significantly change after reoxygenation. Thus gram-negative bacteremic stimulation of the liver or brief hypoxia were equally potent in activating NF-B.

View larger version (39K): [in this window] [in a new window]   Fig. 1.   Sequential changes in nuclear factor (NF)-B gel shift pattern over 180 min after Escherichia coli (EC) infection or 0.9% NaCl (NS) challenge. Constant-flow perfusion under normoxic conditions is compared with similar challenge after 30 min hypoxia (H) at t = 0 followed by 120 min of reoxygenation (H/R) (EC + H/R, NS + H/R). A: electrophoretic mobility shift assays (EMSAs) specific for NF-B were performed as described in the MATERIALS AND METHODS. B: means ± SE densitometric analysis of EMSAs specific for NF-B for 3-5 livers in each group are expressed as a percentage of the EC control group.  P < 0.01 compared with normoxic EC control group at t = 60 min.

Xanthine oxidase inhibition reverses hypoxic suppression of postbacteremic NF-B activation. Because intraportal gram-negative bacteremic infection and nonseptic hypoxia similarly increased NF-B activation, whereas the sequential combination of these stimuli decreased its activity, we next examined the effects of xanthine oxidase-derived ROS on the activation of NF-B after these challenges (Fig. 2A). Xanthine oxidase inhibition in normoxic EC livers (Allo + EC group) caused no change in NF-B activation compared with EC controls. However, such Allo pretreatment completely abrogated the hypoxic suppression of postbacteremic NF-B activation in Allo + EC + H/R livers. In support of these results, livers from animals fed a tungsten-enriched diet to inactivate xanthine oxidase activity had directionally similar effects on hypoxic inhibition of postbacteremic NF-B activity as did Allo pretreatment (data not shown). Figure 2B shows the averaged densitometry of each Allo experimental group as a percentage of normoxic EC control livers at t = 60 min. Pretreatment with Allo (Allo + EC) did not significantly alter the activation of NF-B compared with EC infection alone (Allo + EC at t = 60 92 ± 23%, Allo + EC at t = 180 108 ± 30% of EC control), but did, however, completely reverse the inhibition seen after postbacteremic hypoxia (Allo + EC + H at t = 60 105 ± 10%, Allo + EC + H/R at t = 180 108 ± 22% of EC control).

View larger version (34K): [in this window] [in a new window]   Fig. 2.   Sequential changes in NF-B gel shift pattern over 180 min after EC infection or xanthine oxidase inhibition by allopurinol (Allo) before EC infection (Allo + EC). Constant-flow perfusion under normoxic conditions is compared with similar challenge after 30 min H at t = 0 followed by 120 min of reoxygenation (H/R) (Allo + EC + H and Allo + EC + H/R). A: EMSAs specific for NF-B were performed as described in MATERIALS AND METHODS. B: means ± SE densitometric analysis and SE of EMSAs specific for NF-B for 2-5 livers in each group are expressed as a percent of the EC control group.  P < 0.01 compared with normoxic EC control group at t = 60 min.

Supershift analysis confirms heterodimeric identity of NF-B. To analyze the specific subunits involved in NF-B nuclear translocation after EC or hypoxia, antibodies specific to p65 or p50 subunits were used in supershift assays. Figure 3A demonstrates that in both the EC and NS + H/R groups, p65 is the predominant subunit with an equal or additive shift after the addition of both p65 and p50 antibodies. In addition, the xanthine oxidase inhibitor Allo had no effect on the postendotoxic subunit supershift either in the presence or absence of secondary hypoxia. To verify NF-B identity, competition reactions with unlabeled NF-B oligonucleotide or noncompetitive oligonucleotide were performed. The specific unlabeled NF-B oligonucleotide inhibited the binding of the radiolabeled probe, whereas no interference was noted after the addition of the noncompetitive oligonucleotide. The minimal NF-B band noted in the t = 0 livers also demonstrated the same supershift and competition patterns as all other groups tested (data not shown).

View larger version (72K): [in this window] [in a new window]   Fig. 3.   Supershift analysis of NF-B nuclear translocation in the EC and NS + H/R groups (A) or Allo + EC and Allo + EC + H/R groups (B) demonstrates maximal shift with p65 or combination of p65/50 antibodies. Excess unlabeled oligonucleotide completely abrogates the observed NF-B band (arrow), whereas the noncompetitive (non-comp) oligonucleotide has no effect on the observed NF-B band (EMSAs performed as described in MATERIALS AND METHODS).

Hypoxic suppression of postbacteremic TNF- production: effects of xanthine oxidase inhibition. Concomitant with the observed hypoxic suppression of NF-B activation in EC + H/R livers (Fig. 1), cell-associated bioactive TNF- concentrations in whole liver lysates were inhibited by hypoxia in EC + H livers compared with normoxic EC control organs (853 ± 107 vs. 4,463 ± 1,134 U/mg, P < 0.01). This hypoxic suppression persisted during the reoxygenation phase in EC + H/R livers (1,371 ± 170 U/mg at t = 180 min; Fig. 4). As for NF-B activation, this O₂-dependent reduction in TNF- protein synthesis was abrogated by xanthine oxidase inhibition in Allo + EC + H/R livers (2,945 ± 491 U/mg, P = NS vs. normoxic EC control values). Immunoreactive TNF- exported by the liver, as assessed by cytokine concentrations, showed parallel directional changes. Values in venous perfusates rose from a baseline of 4.2 ± 2.8 to 17.8 ± 1.8 mg/ml after EC infection and were significantly reduced after sequential EC + H/R (2.8 ± 0.4 mg/ml; P < 0.01) (Fig. 5). However, inhibition of xanthine oxidase-derived ROS by Allo did not offset the H/R-mediated reduction in secreted immunoreactive TNF- (6.3 ± 1.5 mg/ml in Allo + EC + H/R vs. 26.2 ± 6.1 mg/ml in Allo + EC; Fig. 5).

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Serial changes in cell-associated bioactive tumor necrosis factor- (TNF-) over 180 min after intraportal infection at t = 0 with 109 viable EC, pretreatment with Allo followed by EC infection, or isovolumetric NS. Constant-flow perfusion under normoxic conditions (EC, NS, Allo + EC) is compared with 30 min H beginning at 0.5 h after challenges at t = 0, followed by 120 min of R (EC + H/R, NS + H/R, Allo + EC + H/R) (see MATERIALS AND METHODS). Data are means ± SE. * P < 0.01 vs. NS controls.  P < 0.01 vs. EC time-matched control.

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Serial changes in hepatic secretion of immunoreactive TNF- over 180 min after intraportal infection at t = 0 with 109 viable EC or pretreatment with Allo followed by EC infection. Constant-flow perfusion under normoxic conditions (EC, Allo + EC) is compared with 30 min H beginning at 0.5 h after challenges at t = 0, followed by 120 min of R(EC + H/R, Allo + EC + H/R) (see MATERIALS AND METHODS). Data are means ± SE. * P < 0.01 vs. t = 0.  P < 0.01 vs. EC or Allo + EC.

Hypoxic-induced suppression of postbacteremic NF-B activation is not due to organ dysfunction. Inhibition of EC-induced NF-B activation by secondary hypoxia was not accounted for by differences in bacterial clearance from circulating perfusates, as equivalent peak cfu per milliliter at t = 30 min in all EC-infected organs showed similar complete clearance of infused inocula by t = 120 min in all groups (data not shown). In addition, the sequential changes in O₂ in normoxic EC and NS controls were similar at baseline in all preparations and stable thereafter (Fig. 6). The 92 ± 2% reductions in hepatic O₂ delivery to EC + H/R and NS + H/R livers from t = 30 to 60 min were accompanied by parallel decreases in O₂ that returned to prehypoxic levels during the reoxygenation phase in all groups (Fig. 6). Livers pretreated with Allo or harvested from tungsten-fed animals had similar O₂ responses during normoxic and hypoxic treatments (data not shown). Oxidative responses to 10⁷ M glucagon, defined as a >5% increase in O₂ (34), were also not impaired even after the 180 min H/R, as a 23 ± 3% rise after glucagon challenge was present in EC + H/R, Allo + EC + H/R, and NS + H/R livers.

View larger version (21K): [in this window] [in a new window]   Fig. 6.   Sequential changes in hepatic O2 consumption (O2) during constant-flow perfusion for 180 min under normoxic conditions or after 30 min H and 120 min of R after intraportal infection at t = 0 (arrow) with 109 viable EC or NS. Values are means ± SE. * P < 0.01 vs. group-specific values at t = 30 min or 60 min and time-matched EC and NS normoxic controls. O2 values in Allo + EC and Allo + EC + H/R groups were similar to time-matched EC and EC + H/R livers (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This study extends our previous findings that combined, sequential intraportal gram-negative infection and secondary hypoxia/reoxygenation downregulate bioactive and immunoreactive TNF- production in the liver by a mechanism involving decreasing cytokine gene transcription via reduced NF-B activation (20, 34). To date, this is the first report concerning the effects of combined EC + H/R on transactivation of NF-B at the intact organ level. Whereas both EC and H/R individually stimulate the activation of NF-B, a brief and otherwise well-tolerated period of postbacteremic hypoxia significantly inhibits the intrahepatic activation of NF-B. This inhibition is not due to differences in bacterial clearance nor to the inability to restore O₂ during reoxygenation.

Such as others have described in cell culture experiments or postmortem studies (3, 9, 33), we noted an increase in NF-B activity and subsequent TNF- production after infection with gram-negative bacteria in ex situ-perfused liver experiments under normoxic conditions. In Kupffer cells, an increase in NF-B activation within 30 min after stimulation with LPS that lasted 24 h was noted (33). This activation was inhibited by the ROS scavenger pyrrolidine dithiocarbamate. In the ex situ rat liver (9), Salmonella enteriditis endotoxin increased NF-B activity within 1 h, and, in the rat lung (3), increased NF-B activity after intraperitoneal infection with EC or Salmonella typhimurium has been demonstrated to occur by 1 h as well. This activation was reduced with the intracellular glutathione-repleting and direct ROS scavenger N-acetylcysteine. Endotoxin regulates NF-B activation and subsequent cytokine gene expression in part via the production of ROS activating tyrosine kinases, although signaling via CD14 receptors that is independent of tyrosine kinase activation may also be important (6, 12). Endotoxin directly effects the intracellular redox status (2, 18); however, most evidence is based on the modulatory effects of antioxidants and suggests that xanthine oxidase-mediated ROS accumulation is of importance for LPS signaling events (3, 18, 31, 33). In the perfused rat liver model employed herein, postbacteremic NF-B activation was not abrogated by Allo nor by tungsten treatments, suggesting that xanthine oxidase-mediated ROS are not implicated in the gram-negative intracellular signaling pathway under normoxic conditions.

There is controversy among reports as to whether hypoxia alone is a sufficient stimulus for NF-B activation (15, 16, 24, 26, 27). This is likely due to differences in experimental design concerning the severity and duration of hypoxia, the variety of model systems, and lack of data at the intact organ level. NF-B was not activated in HeLa cells exposed to <10 mmHg O₂ for up to 5 h, although activation occurred promptly with reoxygenation (26). In contrast, a period of anoxia activated NF-B within 3 h in Jurkat T-, NIH3T3, and mouse L-TK cells, although the nuclear translocation was slower than that seen after stimulation with TNF-, phorbol 12-myristate 13-acetate, or IL- (15, 16, 24). There is one report of a more modest reduction in O₂ (PO₂ ~40 mmHg) activating NF-B in human umbilical vein endothelial cells (27). Our data confirm and extend these findings and are the first to examine the effects of modest hypoxia on NF-B activation at the whole organ level. We presume that these effects have been due to functional changes in the Kupffer cells; however, we cannot exclude the possibility that the effects of hypoxia may be mediated indirectly through changes in the function of the hepatocytes or other liver cell subtypes. Using a reduction in the hepatic O₂ delivery likely to occur early after human bacteremic sepsis and one that caused no evident cellular damage, we demonstrate that hypoxia alone, as modeled in the NS + H/R livers, is a sufficient and potent stimulus for the activation of NF-B within the intact liver.

To date, no other study has examined changes in transcription factor activation after the combined, sequential oxidant stresses of gram-negative bacteremic infection and modest secondary hypoxia. As both bacterial products and reductions in cellular oxygenation have been shown to independently activate NF-B (1), the mechanism(s) whereby secondary hypoxia inhibits EC-induced activation of this transcription factor remain unclear. Although the membrane signaling events after H/R appear to differ from those after cytokine or LPS stimulation, downstream effector mechanisms involving tyrosine kinase activation appear similar, albeit with differing activation kinetics (12, 15-17). For example, anoxia as well as TNF- or phorbol esters activate Ras and Raf tyrosine kinases that phosphorylate IB, leaving this natural inhibitor of NF-B susceptible to degradation via serine protease activity (15, 16). LPS may also activate NF-B via another intracellular pathway that is not tyrosine kinase dependent (6). Thus the combination of endotoxin and secondary hypoxia may compete for kinase substrate or may alter the cytoplasmic balance between kinase and phosphatase activity, interfering with signal propagation and subsequent activation.

A more likely explanation for the hypoxic inhibition of postbacteremic NF-B activation is the augmented oxidant stress after the combined stimuli. The highly conserved Rel homology domain contains a cysteine residue that must be maintained in a reduced state for maximal NF-B binding to its cognate DNA sequences (13, 19). Hence, the intracellular redox milieu as reflected by reduced vs. oxidized thiol availability appears to dynamically regulate both NF-B activation and its binding to DNA consensus sites within the TNF- promoter region. We postulate that the combined production of ROS due to combined, sequential intraportal infection and secondary hypoxia and reoxygenation creates an intracellular milieu in which translocated NF-B binding to cognate DNA is inhibited. The posthypoxic inhibition of endotoxin-induced NF-B activation was reversed by pretreatment with either Allo or tungsten, implicating the xanthine oxidase-mediated ROS generation as being central in the signaling pathway leading to NF-B activation and TNF- production after hypoxia. In support, other studies have also suggested that the xanthine oxidase pathway is the major source for ROS production during postanoxic reoxygenation (14, 36). In this context, the ratio between reduced and oxidized glutathione (GSH/GSSG) and other intracellular thiols has been implicated in the function or regulation of diverse immunological functions, including NF-B activation (8, 13, 30). Whereas a minimum level of GSH oxidation appears necessary for NF-B activation and nuclear translocation, excessive intracellular GSSG has been reported to interfere with NF-B binding to DNA (11, 13). We found no differences in intrahepatic GSH levels among any experimental group. This suggests that the threshold changes in the GSH/GSSG that stimulate NF-B activation or inhibit NF-B binding to DNA are small and nonlinear. Alternatively, other thiol donors such as thioredoxin may play a role in this model system.

In this report, we extend our previous findings that TNF- gene expression is strongly inhibited after sequential intraportal EC infection and subsequent H/R in EC + H/R rat livers. Our previous studies (34) demonstrated that, although released bioactive and immunoreactive TNF- were increased after a period of nonbacteremic hypoxia, TNF- mRNA after NS + H/R or EC + H/R was decreased relative to baseline. We demonstrate here that intracellular levels of immunoreactive TNF- are also not increased above baseline by the period of nonbacteremic hypoxia. Taken together, these data suggest that hypoxia may act differentially at transcriptional and translational levels. In support, others have reported that ROS of mitochondrial origin during H/R act at a posttranslational level to release TNF- from the membrane fraction (14). In parallel with NF-B activation, cell-associated TNF- concentrations were nearly fully restored with xanthine oxidase inhibition, further suggesting a role for xanthine oxidase-mediated ROS in the signaling pathway used by hypoxia. Perfusate levels of immunoreactive TNF- were not restored by xanthine oxidase inhibition, in agreement with our earlier findings for perfusate bioactive TNF- levels in the Allo + EC + H/R livers (34). This is likely explained by the fact that the cell-associated TNF- fraction is much larger than the excreted fraction, and alterations in gene production would be expected to be evident in the intracellular pool first. Although NF-B was increased by nonbacteremic hypoxia, the period of hypoxia may have also altered the binding of this transcription factor to one or more of its cognate binding sites within the TNF- promoter or altered other important transcriptional regulators of TNF- gene expression.

Activation of NF-B early after gram-negative bacterial infections or after reduced tissue oxygenation is of pathophysiological significance given the discovery of -B binding sites within the promoter regions of genes that amplify sepsis-related inflammatory responses, including TNF- and IL-1 (4, 28). During critical illnesses, episodes of gram-negative bacterial infections with varying degrees of tissue hypoxia are common. The finding that postbacteremic hypoxia inhibits rather than augments the activation of NF-B under dynamic conditions of perfusion at the organ level may be of clinical importance. The O₂ dependency of bacteremic-induced transcription factor activation and subsequent cytokine production may be an intrinsic homeostatic strategy of the bacteremic host to limit inflammation or, alternatively, may impair cytokine-dependent host defense. Adding to the complexity of redox-dependent transcription factor activation and cytokine expression after sequential bacteremia and H/R is the finding of different organ-specific responses between the liver and the lungs. We previously reported that in the ex situ liver, a period of postbacteremic hypoxia downregulates TNF-, IL-1, and IL-1 (20, 35), whereas postbacteremic alveolar hypoxia upregulates TNF- and IL-1 expression in perfused rat lung (21). In light of these considerations and the time constraints imposed by organ perfusion studies, the results presented herein need to be interpreted cautiously. Further investigations regarding the timing and duration of hypoxia relative to bacteremia and cell- and organ-specific responses will be important to fully define the consequences of these combined stimuli on NF-B- and cytokine-dependent inflammation.