Higher Endotoxin Levels Research Articles

Platelet-activating factor in hepatic ischemia/reperfusion injury. The effects of a PAF antagonist combined with a prostaglandin I2 analogue.

The effects of TCV-309, a specific platelet-activating factor (PAF) antagonist, and OP-41483, a prostaglandin I2 analogue, on warm ischemia/reperfusion injury of the rat liver were studied. Rats were divided into five groups by the duration of warm ischemia and the treatment used. The NS1 group (normal saline pretreatment) had 60 min of warm ischemia, while the NS2 group (normal saline pretreatment), the PGI2 group (OP-41483, 500 ng/kg/min pretreatment), the TCV group (TCV-309, 3 micrograms/kg), and the PGI2+TCV group (both the above dosages) underwent 120 min of warm ischemia. Postoperative survival after 30 days, bile secretion, serum endotoxin levels, and tissue glutathione levels after 60 min of reperfusion were compared between the groups. The survival rates for the NS1, NS2, PGI2, TCV, and PGI2+TCV groups were 80%, 0%, 50%, 80%, and 86.7%, respectively. Bile secretion, which has a strong correlationship with hepatic cellular ATP level, was strongly correlated with survival. The NS2 group had a high serum endotoxin level--however, the PGI2 and PGI2+TCV groups had normal levels. Although there were some discrepancies between survival and the tissue glutathione level, combined treatment with the PGI2 analogue and TCV-309 was most effective in inhibited oxidative stress. In conclusion, TCV-309 increased the survival rate after 120 min of warm hepatic ischemia without endotoxemia by the PGI2 analogue. This finding suggest that warm ischemia/reperfusion injury is related to the generation of PAF. Combined pretreatment with TCV-309 and a PGI2 analogue may be useful in liver transplantation.

The release of endotoxin from antibiotic-treated Escherichia coli and the production of tumour necrosis factor by human monocytes.

The influence of antibiotic-induced release of endotoxin from in-vitro grown Escherichia coli on the production of tumour necrosis factor-alpha (TNF) by human monocytes was studied. Antibiotics tested were: cefuroxime (7.5 and 75 mg/L); ceftazidime (10 and 100 mg/L); aztreonam (10 and 100 mg/L); imipenem (10 and 100 mg/L); and tobramycin (8 mg/L). The effect of the combination of cefuroxime plus tobramycin, and the effect of taurolidine, an endotoxin-binding agent, on TNF production was also tested. After incubation for 4 h, all antibiotic-treated cultures (high-dose) induced a similar rise in extracellular TNF production when compared to the controls. However, after incubation for 24 h, a significant rise in TNF production was noticed in the cefuroxime and aztreonam-treated cultures (6440 and 5969 ng/L, respectively) compared to the ceftazidime and imipenem-treated cultures (846 and 381 ng/L, respectively). The cefuroxime-induced release of TNF could be reduced by addition of tobramycin (from 6440 to 1615 ng/L). Similar differences in TNF production were noticed in cell-associated TNF. Dose-response curves did not demonstrate differences in TNF production in aztreonam or imipenem-treated cultures. However, for both cefuroxime and ceftazidime-treated cultures, low-dose treatment resulted in significantly higher production of TNF. The differences in TNF production between these antibiotics could be explained by the production of filaments following treatment with cefuroxime, aztreonam and low-dose ceftazidime, resulting in late bacterial lysis with high levels of endotoxin, whereas treatment with imipenem or high-dose ceftazidime resulted in the formation of spheroplasts, resulting in early lysis of the bacteria and much lower levels of endotoxin. The addition of taurolidine to either imipenem or aztreonam-treated cultures prevented a rise in TNF production as a result of nearly complete neutralization of the released endotoxin. It was concluded that the observed differences in TNF production by human monocytes in vitro were related to differences in the mechanisms and amount of antibiotic-induced release of endotoxin.

Process development for the recovery and purification of recombinant protein G

The domains of protein G from streptococcus which bind immunoglobulin G have been cloned and expressed in Escherichia coli (Fahnestock et al., 1986). Because protein G binds to several animal immunoglobulin G's, it has many immunochemical applications. This report describes process development for large-scale production of this recombinant protein G (also known as GammaBind® G). In 200 l cultures of E. coli, this protein G variant was released from the cell into the culture medium by heating at 80°C for 10 min. The concentration was monitored by either a competitive enzyme-linked immunoassay or a liquid chromatographic assay. Cross-flow microfiltration with 0.22 μm membrane was used to remove the cells. The protein G-rich permeate from the crossflow microfilter was purified by affinity chromatography using a 5 l column of IgG-Sepharose 6 Fast Flow, which yielded 16–18 g of protein G per column cycle. The pools of purified protein G were concentrated and desalted using ultrafiltration. The salt-free protein G was then lyophilized as bulk product. The overall recovery through the entire process was 50–64%. The analysis of the final product included sodium dodecyl sulfate polyacrylamide gel electrophoresis, UV-visible spectrum, high performance gel filtration, endotoxin level and binding efficiency to human IgG Sepharose.

Evidence against oxidant injury and endotoxin underlying glycerol-induced fatal rhabdomyolysis in rats.

Rhabdomyolysis in experimental animals and in man may be fatal. The mechanism of action causing death in man and in the experimental model of laboratory animals injected with glycerol in a dose causing fatal rhabdomyolysis is unknown. The aim of this study was to examine two possible causes of death following glycerol injection (0.5-2 ml/100 g) to the leg muscle of rats following a period of water deprivation (24-72 h). The first factor examined was free radical formation, via the administration of various antioxidants such as natural antioxidant (NAO - 5 mg/100 g), vitamin E (0.2 mg/100 g each day for 7 days), dimethylthiourea (DMTU - 50 mg/100 g) and superoxide dismutase (SOD - 0.8 mg/100 g). The second factor was a high blood endotoxin level due to the glycerol injection. This possibility was examined by glycerol injection to rats that had developed endotoxin tolerance. The low survival rate (0-30%) of rats receiving a glycerol injection following preventive treatment with antioxidants as well as rats which had developed endotoxin tolerance (0-40%), excludes the possibility that the cause of death in the glycerol model resulting in fatal rhabdomyolysis is due to free radical formation or to a high and unneutralized blood level of endotoxin.

High serum IL-6 level reflects susceptible status of the host to endotoxin and IL-1/tumor necrosis factor.

Patients with high level of serum endotoxin did not necessarily develop into lethal shock, whereas some patients died of septic shock even when their serum endotoxin levels were low. These results indicate that limiting factor which determines the host to be endotoxin shock principally depends on the host susceptibility to endotoxin instead of serum endotoxin level. To understand this susceptible status of the host to endotoxin, we used Propionibacterium acnes primed mouse endotoxin shock model. We found that P. acnes-primed mice responded to low dose of LPS by enhanced production of IL-1 and TNF. And such mice were highly susceptible to the lethal shock inducing effect of IL-1 and/or TNF, which also induced high level of serum IL-6 in these mice. Therefore, measurement of serum IL-6 level provides us with the information of the preceding exposure of the host to either LPS or IL-1 and/or TNF and the highly susceptible status of the host to these stimuli. Based on these results obtained from animal model, we investigated the relationship between serum IL-6 levels and serum endotoxin levels in the patients with malignant hematologic disorders. We found that these patients fell into two groups; an endotoxin susceptible group, equivalent to P. acnes-primed mice, showing high level of serum IL-6 with low level of serum endotoxin, and a nonendotoxin susceptible group, equivalent to P. acnes-nonprimed mice, showing low or undetectable level of serum IL-6 with high level of serum endotoxin. We propose that the measurement of serum IL-6 level in the patients positive for endotoxin is a useful tool in evaluating diagnosis and prognosis of endotoxin shock.

Vaccination of Turkeys with Cell-Free Culture Filtrate of Pasteurella multocida: Effects of Ultrafiltration and Endotoxin Removal

Cell-free culture filtrate (CCF) of Pasteurella multocida strain R44/6 (serotype 3/4/9/12) was fractionated by ultrafiltration into fractions of less than 10,000, greater than 10,000, greater than 30,000, and 10,000 to 30,000 molecular weight (MW). The less-than-10,000-MW fraction contained little endotoxin comparable to bacteriologic medium; the 10,000-to-30,000-MW fraction had a moderate amount of endotoxin, whereas the greater-than-10,000- and greater-than-30,000-MW fractions contained high levels of endotoxin. Following ultrafiltration, each fraction, except the less-than-10,000-MW fraction, was divided into two equal parts, and endotoxin was removed from one part. Turkeys were vaccinated with the various MW fractions of CCF, with and without endotoxin, via the air sacs at 6 and 9 weeks of age and compared with negative controls given bacteriologic medium and positive controls vaccinated with a commercial bacterin. Before oral challenge with strain P-1059 (serotype 3) at 12 weeks of age, antibody titers were detected only in positive control turkeys. Protection against challenge, as measured by post-challenge mortality and body-weight gain, was provided by the greater-than-10,000-, greater-than-30,000-, and 10,000-to-30,000-MW fractions containing endotoxin and the commercial bacterin. Turkeys that had been vaccinated with bacteriologic medium and the four different fractions without endotoxin were not protected. Results indicated that endotoxin in CCF of P. multocida is critical in protecting turkeys from pasteurellosis.

Nitrogen Oxide Levels in Patients After Trauma and During Sepsis

The mediators responsible for maintenance of the hyperdynamic state and the low systemic vascular resistance (SVR) observed in sepsis have not been elucidated. Nitric oxide (.N = O) is a mediator with numerous functions, including regulation of vascular tone and a role in macrophage-mediated cytostasis and microbiostasis. Thirty-nine critically ill trauma and septic patients were studied to determine the relationship between .N = O production and the hyperdynamic state. high plasma levels of NO2-/NO3- (the stable end products of .N = O) were observed in septic patients (p less than 0.02). Low SVR and high endotoxin levels were associated with high NO2-/NO3- values (p = 0.029, p = 0.002). Changes in .N = O levels may mediate the vasodilation seen in sepsis. Low NO2-/NO3- levels were observed in trauma patients (p less than 0.001) and remained low even in the presence of sepsis (p = 0.001).

Filtration of Dialysate Using An On-Line Dialysate Filter

Increased concerns about pyrogenic contamination of dialysate have led to the development of an on-line dialysate filtration system. Bacteriological testing of the system was performed (n = 6) by introducing bicarbonate concentrate contaminated with E. coli 026:B 6 (3 x 10(9) cfu/ml) into a dialysis machine equipped with a two-stage polysulfone filtration system. The bacterial concentration of the dialysate entering the filtration system was maintained above 10(6) cfu/ml and endotoxin levels ranged from 30-300 ng/ml during the 3-hour test period. Bacterial and endotoxin levels on the input side of the first-stage filter reached minimum concentrations of 5.4 x 10(9) cfu/ml and 30,000 ng/ml respectively. All output samples of filtered dialysate showed no bacterial growth and endotoxin levels were below the sensitivity (0.003 ng/ml) of the LAL assay. A dialysis machine (QD = 500), equipped with a single stage filtration system, was used for 18 months of clinical testing. In order to evaluate the system's reliability with regard to membrane failures and reduced dialysate flow, filter membrane integrity was verified weekly using a pressure holding test and dialysate flow was measured under routine clinical conditions. No membrane failures occurred, and dialysate flow was maintained at 511 +/- 17 ml/min (n = 70) during the test period. dialysate filtration is an effective and practical method for prevention of pyrogenic reactions due to high levels of bacteria and endotoxins.

Staphylococcus epidermidis induces complement activation, tumor necrosis factor and interleukin-1, a shock-like state and tissue injury in rabbits without endotoxemia. Comparison to Escherichia coli.

Tumor necrosis factor (TNF) and IL-1 are thought to mediate many of the pathophysiologic changes of endotoxemia and Gram-negative bacteremia. In these studies, heat-killed Staphylococcus epidermidis were infused into rabbits to determine whether an endotoxin (LPS)-free microorganism also elicits cytokinemia and the physiologic abnormalities seen in Gram-negative bacteremia. S. epidermidis induced complement activation, circulating TNF and IL-1, and hypotension to the same degree as did one-twentieth the number of heat-killed Escherichia coli. Circulating IL-1 beta levels had a greater correlation coefficient (r = 0.81, P less than 0.001) with the degree of hypotension than TNF levels (r = 0.48, P less than 0.02). Leukopenia, thrombocytopenia, diffuse pulmonary capillary aggregation of neutrophils, and hepatic necrosis with neutrophil infiltration were observed to the same extent after either S. epidermidis or E. coli infusion. However, S. epidermidis infusion did not induce significant (less than 60 pg/ml) endotoxemia, whereas E. coli infusion resulted in high (11,000 pg/ml) serum endotoxin levels. S. epidermidis, E. coli, LPS, or S. epidermidis-derived lipoteichoic acid (LTA) induced TNF and IL-1 from blood mononuclear cells in vitro. E. coli organisms and LPS were at least 100-fold more potent than S. epidermidis or LTA. Thus, a shock-like state with similar levels of complement activation as well as circulating levels of IL-1 and TNF were observed following either S. epidermidis or E. coli. These data provide further evidence that host factors such as IL-1 and TNF are common mediators of the septic shock syndrome regardless of the organism.

Removal of Endotoxin and Cytokines by Adsorbents and the Effect of Plasma Protein Binding

High serum levels of endotoxin and cytokines, through which its activity is mediated, have been shown to be associated with disease severity in septic shock and in fulminant hepatic failure. In the present study, we have investigated the ability of activated charcoals (DHP-1 and Adsorba 150C) and uncharged resin (Amberlite XAD-7) to adsorb lipopolysaccharide (LPS) and various cytokines, namely tumour necrosis factor (TNF), interleukin-1 (IL-1), interleukin-6 (IL-6), interferon-alpha (IFN-alpha) and interferon-gamma (IFN-gamma). The capacities of the adsorbents were assessed by measurement of their equilibrium adsorption isotherms for these substances labelled with 125I. There was no single adsorbent that uniformly adsorbed LPS and the cytokines from phosphate buffered saline or human plasma. DHP-1 charcoal was superior to Adsorba 150C for all substances and was the most effective adsorbent for binding LPS, IL-1 alpha and IFN-gamma. Amberlite XAD-7 resin was most effective for TNF, IL-6 and IFN-alpha, but bound little LPS, particularly from human plasma. Ultrafiltration through a membrane which retains substances of molecular weight greater than 50 kD did not filter the cytokines from human plasma, although the molecular weight of the cytokines range from 17 to 22 kD. This demonstrated that, TNF, IL-1, IL-6, IFN-alpha and IFN-gamma readily bind to proteins and/or other large molecules in plasma.

In vitro Study of the Transfer of Cytokine-Inducing Substances across Selected High-Flux Hemodialysis Membranes

The in vitro transfer of cytokine-inducing substances (CIS) across cellulose triacetate and polyacrylonitrile hollow-fiber high-flux hemodialyzers was studied using culture filtrates of gram-negative bacteria isolated from hemodialysis center environments. With Enterobacter cloacae, no transfer of CIS was seen despite the potent cytokine inducibility and endotoxin content of the challenge solution. In contrast, interleukins 1 and 6 and tumor necrosis factor inducing substances did penetrate both dialyzer types challenged with Pseudomonas aeruginosa culture filtrates containing a high endotoxin content. Transfer was not seen, however, upon dilution of the challenge solution to lower, yet clinically very high levels of endotoxin. These results show that, in vitro, the transfer of CIS across high-flux membranes is critically dependent upon the quality and the quantity of the challenge material employed.

Removal of Endotoxin and Cytokines by Adsorbents and the Effect of Plasma Protein Binding

High serum levels of endotoxin and cytokines, through which its activity is mediated, have been shown to be associated with disease severity in septic shock and in fulminant hepatic failure. In the present study, we have investigated the ability of activated charcoals (DHP-1 and Adsorba 150C) and uncharged resin (Amberlite XAD-7) to adsorb lipopolysaccharide (LPS) and various cytokines, namely tumour necrosis factor (TNF), interleukin-1 (IL-1), interleukin-6 (IL-6), interferon-alpha (IFN-alpha) and interferon-gamma (IFN-gamma). The capacities of the adsorbents were assessed by measurement of their equilibrium adsorption isotherms for these substances labelled with 125I. There was no single adsorbent that uniformly adsorbed LPS and the cytokines from phosphate buffered saline or human plasma. DHP-1 charcoal was superior to Adsorba 150C for all substances and was the most effective adsorbent for binding LPS, IL-1 alpha and IFN-gamma. Amberlite XAD-7 resin was most effective for TNF, IL-6 and IFN-alpha, but bound little LPS, particularly from human plasma. Ultrafiltration through a membrane which retains substances of molecular weight greater than 50 kD did not filter the cytokines from human plasma, although the molecular weight of the cytokines range from 17 to 22 kD. This demonstrated that, TNF, IL-1, IL-6, IFN-alpha and IFN-gamma readily bind to proteins and/or other large molecules in plasma.

Endotoxin and Bacterial Contamination of Dialysis Center Water and Dialysate; a Cross Sectional Survey

The bacterial and endotoxin levels of purified water and effluent dialysate were examined in a cross section of dialysis centers in the central United States. All samples were collected within a four-hour drive of the University of Louisville and were collected, processed and analyzed by our personnel, to eliminate variability in sample handling. A medium capable of higher bacteria recovery from aqueous environments than those ordinarily employed in clinical assays was used. Endotoxins were determined by a quantitative colorimetric assay. By the more sensitive bacterial assay 53% of the centers had bacterial counts above the AAMI standard of 200 colony-forming units per ml (CFU/ml) for water and 35% of the centers had bacterial counts above the 2000 CFU/ml standard for dialysate in at least one sampling period. The samples showed 35% and 19% of water and dialysate above the standards, respectively. While there are no standards for endotoxin concentrations in water used to prepare dialysate, 2% of the centers had endotoxin levels in their water above five endotoxin units per ml (5 EU/ml = 1 ng/ml in our assay kit), the limit set by the AAMI standards for reprocessor water. Both bacterial and endotoxin levels tended to be elevated in dialysate, with the highest levels of endotoxin in dialysates posing an obvious potential risk when high-flux dialyzers are used.

Endotoxemia in Patients on Hemodialysis

We examined endotoxins and limulus amebocyte lysate-reactive material (LAL-RM) in serum from 87 patients on hemodialysis, using the conventional chromogenic limulus test (CCLT) and a new specific endotoxin assay with factor G-free LAL (endotoxin-specific test: EST). All patients with regenerated cellulose dialyzers had increased CCLT values, whereas the EST values were not higher than in controls. This discrepancy can be explained by the LAL-RM which is cellulose-derived and cross-reacts with LAL via factor G. Using EST for measurements of endotoxin, 6 patients out of 87 had pathological endotoxemia and all of these patients were associated with either cirrhosis, infection, or malignancy. Some patients who had no complications showed fever before or during dialysis, but they did not have high endotoxin levels. EST is useful for the diagnosis of endotoxemia in patients on hemodialysis because of the presence of LAL-RM in serum. Endotoxemia is rare in patients on hemodialysis, if they are not associated with infection, cirrhosis, or other complications.

Outbreak of Gram-Negative Bacteremia and Pyrogenic Reactions in a Hemodialysis Center

During the period from April 4, 1988, to April 20, 1988, nine pyrogenic reactions and five gram-negative bacteremias occurred in 11 patients undergoing dialysis. All pyrogenic reactions and gram-negative bacteremias occurred among patients in whom a reprocessed dialyzer was used. The rate of pyrogenic reactions or bacteremias per 100 sessions using a reprocessed dialyzer was higher than in sessions during which a new dialyzer was used (4.5 vs. 0; p = 0.03). Dialyzers were manually reprocessed with 2.5% Renalin germicide. The Renalin concentrations varied widely in 12 dialyzers stored after manual reprocessing during the epidemic period (0.9-4.2%); the median endotoxin concentrations varied from 0 to 246 ng/ml and were higher in dialyzers with Renalin concentrations less than or equal to 1.0% than in dialyzers with higher concentrations (p = 0.01). Experiments using a dilution technique described by a technician resulted in Renalin concentrations ranging from 1.4% at the surface to 3.5% at the bottom of the preparation container. These findings suggest that failure to adequately admix Renalin during dilution may be associated with low levels of disinfectant, high levels of bacteria and endotoxins in dialyzers, and outbreaks of pyrogenic reactions and gram-negative bacteremias in dialysis patients.

Plasminogen activator inhibitor 1 and 2, alpha-2-antiplasmin, plasminogen, and endotoxin levels in systemic meningococcal disease

We have studied the activation state of the fibrinolytic system in 39 patients with systemic meningococcal disease (SMD). Patients defined as having fulminant septicemia (n=13) with high (>700 ng/L) levels of endotoxin (LPS) in plasma and severe coagulopathy, had significantly lower functional levels of plasminogen (P< 0.05) and alpha-2-antiplasmin (P > 0.01) and higher antigen levels of plasminogen activator inhibitor 1 (PAI-1) (P <0.01), and fibrin degradation products (FDP) (P< 0.01), but not of PAI-2 (P >0.1) as compared with less severely ill patients (meningitis and meningococcemia)(n=25). A positive correlation existed between the admission (maximum) levels of LPS and PAI-1 (r= 0.86, P <0.0001). Decreasing admission levels of platelets were associated with increasing levels of PAI-1 (r=-0.55, P<0.001). After initiation of treatment with antibiotics and fresh frozen plasma, the PAI-1 levels declined rapidly. PAI-1 levels >360 ug/L on admission predicted the development of a severe septic shock combined with renal impairment correctly in 12 of 13 patients (92%). None of 25 patients without multiple organ failure had PAI-1 levels >260 ug/L. PAI-1 levels >1850 ug/L were associated with 100% fatality. The results suggest that in the early phase of fulminant meningococcal septicemia an extensive plasmin generation occurs. On admission, however, high levels of PAI-1 seem to inhibit the plasmin generation, and thereby promote DIC .

Detection of endotoxin in cases of equine colic.

The Limulus amoebocyte lysate assay was used to test for the presence of endotoxin in 37 clinical cases of equine colic. Positive plasma titres were detected in 10 cases and the presence of endotoxin was significantly correlated with a high heart rate, a high packed cell volume and a poor prognosis. High levels of endotoxin were detected in gut contents taken from several sites in the gastrointestinal tract of normal horses.

Endotoxin levels measured by a chromogenic assay in portal, hepatic and peripheral venous blood in patients with cirrhosis.

Endotoxin concentrations were measured in the portal, hepatic and peripheral venous blood of two groups of patients with cirrhosis using a limulus-based chromogenic assay. The high sensitivity of chromogenic detection allowed measurement of endotoxin as low as 10 to 15 pg per ml, an order of magnitude greater than previously possible by gelation studies. Group 1 consisted of 56 patients with cirrhosis undergoing angiographic evaluation. In this group, there was wide variability in hepatic venous concentration [73 +/- 110 pg per ml (mean +/- S.D.)] and peripheral venous concentration [31 +/- 58 pg per ml]. However, paired t test showed peripheral venous concentration was significantly (p less than 0.001) lower than hepatic venous concentration. Neither hepatic or peripheral venous endotoxin levels correlated significantly with a variety of clinical, biochemical or radiological parameters. Group 2 consisted of 21 patients with cirrhosis undergoing shunt surgery. Endotoxin levels again showed a wide range, with portal venous concentration (142 +/- 167 pg per ml) and simultaneous peripheral venous concentration (82 +/- 150 pg per ml). Paired t test in this group showed a significant (p less than 0.001) portal to peripheral venous gradient. This study showed the feasibility of measuring endotoxin in plasma to low concentrations by a chromogenic assay technique. It supports the concept of relatively high levels of endotoxin in the portal circulation. In the presence of liver disease, systemic endotoxemia occurs, which is augmented by stressful situations.

3-Hydroxymyristic acid as a measure of endotoxin in cotton lint and dust.

A series of samples consisting of purified cellulose, purified cellulose spiked with endotoxin, and cotton lint and dust samples from the Human Panel Acute Exposure Studies at Clemson, South Carolina, were extracted with pyrogen-free water and with phenol-water. Phenacyl esters of the dried, hydrolyzed extract were prepared and chromatographed on a high performance liquid chromatograph. A peak assigned to the phenacyl ester of 3-hydroxymyristic acid appears in the chromatograms of extracts of celluloses that have been spiked with endotoxins and not in those of unspiked celluloses. This peak also appears in the extracts of cotton lint from samples that cause the greatest decrement in lung function in the Clemson human exposure studies. The area of this peak increases with increasing amounts of endotoxin and may serve as a measure of endotoxin concentration in cotton lint and dust, at least when fairly high levels of endotoxin (0.50 micrograms or greater) are present. The effect of extraction method on the determined amount of endotoxin is discussed.

Quantitative measurement of endotoxin in rainbow trout (Salmo gairdneri) serum by the chromogenic substrate method.

The amount of endotoxin in serum collected from normal rainbow trout (Salmo gairdneri) and trout inoculated with viable Vibrio anguillarum or lipopolysaccharide (LPS) extracted from bacteria was determined by the chromogenic substrate method. The mean values of endotoxin in four different groups of normal rainbow trout sera ranged from 31.9 to 65.3 pg/ml. When fish were inoculated with viable bacteria (1 x 10(8], they became septicaemic and a large amount of endotoxin ng/ml) was detected in the sera. In fish inoculated with a smaller number of bacteria the amount of endotoxin was several times higher than that of normal fish in spite of failure of bacterial isolation. Although the endotoxin level in serum increased rapidly (greater than 100 ng/ml) after intraperitoneal inoculation with purified V. anguillarum LPS (540 micrograms), no fish died during the experiment. The high level of endotoxin in normal rainbow trout and the resistance of trout to endotoxin are in striking contrast to those of mammalian and avian species.