INFECrION AND IMMUNITY, Aug. 1993, p. 3143-31480019-9567/93/083143-06$02.00/0Copyright C) 1993, American Society for Microbiology

Alterations of Bacterial Clearance Induced by Endotoxinand Tumor Necrosis Factor

THEA KOCH,lt* HANS P. DUNCKER,1 ROLAND AXT,2 HANS G. SCHIEFER,3KLAUS vA ACKERN,1 AND HEINZ NEUHOF2

Department ofAnesthesiology and Operative Intensive Care, Faculty of Clinical Medicine Mannheim,University ofHeidelberg, Theodor Kutzer Ufer, D-6800 Mannheim, 1 and Division of Clinical Pathophysiolo' y

and Experimental Medicine, Department of Internal Medicine, 2 and Department ofMedical Microbiology,University of Giessen, D-6300 Giessen, Germany

Received 15 January 1993/Accepted 3 May 1993

The purpose of the study was to investigate the potential influence of endotoxin and tumor necrosis factor(TNF) on immune function in terms of systemic clearance and organ distribution of injected Escherichia coli in

a rabbit model. To enable quantification of the clearance process, defined numbers of exogenous E. coli (1.3 x108 CFU) were injected intravenously 60 min after bolus application of TNF (4 x 10' U, n = 6), after infusionof endotoxin (40 ug/kg of body weight) for 1 h (n = 6) or 4 h (n = 6), or after saline infusion (controls, n = 6).Parameters monitored were arterial pressure, oxygen uptake, and rates of bacterial elimination from the blood.At 180 min after E. coli injection, the animals were sacrificed, and tissue samples of liver, kidney, spleen, andlung were collected for bacterial counts. Endotoxin infusion produced a significant delay in blood clearancecompared with saline and TNF pretreatment. The diminished systemic bacterial elimination was associated withsignificantly higher numbers ofE. coli in the organs, thus reflecting reticuloendothelial system dysfunction. TNFhad no major influence on the elimination kinetics of bacteria but affected the tissue distribution pattern withincreased accumulation of E. coli in the lung (up to 100-fold of control values; P < 0.001).

Septic shock and multiple-organ failure continue to bemajor causes of death in severely injured patients (5). That atransient but profound depression of the reticuloendothelialsystem (RES) as well as intestinal barrier dysfunction hasbeen observed after injury or surgical trauma suggests that itplays an essential role in the pathogenesis of septic events.Recent studies (11, 33) documenting an increased permeabil-ity of the gut under trauma and shock conditions, withspreading of endotoxin and bacteria into the circulatingblood and translocation into other organs, emphasized therole of gut barrier failure. The reported pathomechanismssuggest that the incidence of septic complications may bepartially caused, on the one hand, by a loss of barrierfunction, with invasion of great amounts of bacteria andendotoxin exceeding the normal capacity of bacterial clear-ance, and, on the other hand, by an impairment of clearancefunctions possibly due to reduced phagocytosis and lysiscapacity of the RES, or by a combination of both.On the basis of the concept that endotoxin and the release

of cytokines, e.g., tumor necrosis factor (TNF) from mac-rophages which promote the inflammatory reaction, affectthe systemic immune response (26), this study was designedto investigate the influence of endotoxemia on blood andorgan clearance of injected Escherichia coli. Apart from thewell-known effects on intestinal barrier function (10, 13), thecentral question addressed is whether endotoxin directly orindirectly influences phagocytosis and lysis capacity of theRES, thus enhancing bacterial growth in blood and tissues.Since TNF has been shown to be of critical importance in the

* Corresponding author.t Present address: Division of Clinical Pathophysiology and

Experimental Medicine, Department of Internal Medicine,Klinikstrasse 36, D-6300 Giessen, Germany.

development of septic shock in animals and humans chal-lenged with endotoxin, further interest was focused on therelative role ofTNF in mediating septic complications due topossible alterations of bacterial elimination. To simulatebacterial invasion from various compartments, such as thegut, the urogenital tract, wounds, or implanted catheters,into the circulating blood and to enable quantification of theclearance process, defined numbers of exogenous E. coli (1.3x 108 CFU) were injected intravenously after pretreatmentwith endotoxin (40 ,g/kg/h) and TNF (4 x 105 U). Theelimination kinetics of exogenous E. coli from the blood andtheir tissue distribution in the liver, spleen, kidney, and lungwere studied.

MATERIALS AND METHODS

Rabbit model. Standard-breed rabbits of either sex weigh-ing between 2,900 and 3,300 g were anesthetized withpentobarbital (60 to 80 mg/kg of body weight) and anticoag-ulated with heparin-sodium (1,000 U/kg of body weight)injected into an ear vein catheter. The animals were placedin a supine position on a tempered preparation table. Afteraseptic placement of a tracheostomy tube, the rabbits weremechanically ventilated with room air (tidal volume, 30 ml;frequency, 30 volumes per min) by means of a Starling pump(B. Braun, Melsungen, Germany) during the whole observa-tion period. A polyvinyl chloride catheter (inside diameter,1.4 mm) was inserted into the right carotid artery for arterialpressure measurements and blood sampling. Anesthesia wasmaintained by injection of small doses (30 mg) of pentobar-bital under hemodynamic monitoring.

Monitoring. Arterial and airway pressures were monitoredon-line via Statham strain gauge transducers connected to aServomed recorder (Hellige, Freiburg, Germany). Themethod described by Neuhof (30) was used to determine

3143

Vol. 61, No. 8

on May 14, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

3144 KOCH ET AL.

oxygen uptake. The difference in oxygen volume betweenexpiration air and inspiration air was continuously measuredin an open system by means of an oxygen analyzer (Oxy-test-S; Hartmann & Braun, Frankfurt, Germany). The oxy-gen uptake was calculated by the following equation andadjusted to standard temperature, pressure, and drynessconditions: 02 uptake (ml/min) = (air flow [ml/min] x 02difference [ml/100 ml])/100. Blood samples were drawn in-termittently for measurements of pH, PaO2, PaCO2, andHCO3 (ABL 330; Radiometer, Copenhagen, Denmark), he-moglobin and 02 saturation (OSM2; Radiometer), hemato-crit (Adams Autokrit centrifuge; Clay-Adams, Inc., NewYork, N.Y.), and lactate (test-combination lactate, fullyenzymatic; Boehringer GmbH, Mannheim, Germany).

Experimental protocol. After a 30-min steady-state periodof stable hemodynamics and oxygen uptake, the rabbitswere randomly assigned to one of four experimental groups.

(i) Control group (n = 6). After a 60-min period duringwhich 0.9% NaCl (0.1 ml/min) was infused, E. coli (1.3 x 1O0CFU) was injected via the ear vein catheter. Arterial bloodwas aseptically extracted for culture just before and afterbacterial injection at 1, 5, 10, 15, 20, 25, 30, 40, 50, and 60min and thereafter at 30-min intervals. Blood gases, hema-tocrit, hemoglobin, and lactate concentrations were deter-mined at hourly intervals. Three hours after bacterial injec-tion, the animals were sacrificed with an overdose ofpentobarbital, and, subsequently, tissue samples of liver,spleen, kidney, and lung were taken under aseptic conditionsfor quantitative bacterial determinations. The procedurefrom the time point of bacterial injection to 3 h later wasidentical in all experimental groups.

(ii) TNF group (n = 6). After the steady-state period,human recombinant TNF (4 x 105 U dissolved in 2 ml ofbuffer) was injected via the ear vein catheter. Sixty minutesafter TNF application, 1.3 x 108 E. coli was administeredintravenously. The same protocol as that described for thecontrol group was followed. To exclude the possibility ofbacterial translocation induced by TNF in the absence ofinjected bacteria, three experiments without bacterial appli-cation were performed.

(iii) Endotoxin 1 group (n = 6). A nonlethal dose of E. coliendotoxin (40 ptg/kg/h), as assessed in pilot studies, wasinfused for 1 h. Subsequently, E. coli (1.3 x 108) wasinjected intravenously. The same protocol was carried outwithout bacterial injection in three additional experiments.

(iv) Endotoxin 2 group (n = 6). Prolonged endotoxemiawas achieved by infusion of E. coli endotoxin (40 ,ug/kg/h)over a 4-h period prior to bacterial injection. Analogous tothe other groups, bacteria (1.3 x 108 E. coli organisms) wereinjected. To exclude the possibility of bacterial translocationbeing induced only by endotoxemia, three additional animalswere subjected to the same protocol without subsequentbacterial injection.

Quantitative microbiology. Immediately after organ collec-tion, cooled blood and tissue samples were prepared forbacterial culture. Blood samples were diluted by serialdilution into defined volumes of sterile saline. One hundredmicroliters of each dilution was plated in duplicate ontocysteine-lactose electrolyte-deficient agar plates as de-scribed by Sandys (36). Aseptically collected organs wereweighed, and representative samples (0.8 to 2 g) of eachorgan were homogenized in a mortar with 5 ml of sterilesaline. Serial dilutions of tissue suspension (50 p,l) wereplated onto cysteine-lactose electrolyte-deficient agar plates.The inoculated plates were incubated at 37°C for 24 h, andbacterial counts, appearing as CFU, were read. The final

bacterial concentrations were calculated as the numbers ofcolonies per milliliter of blood and as colonies per gram ofharvested tissue.

Materials. Recombinant human TNF was a generous giftfrom Knoll AG (Ludwigshafen, Germany). The proteinconcentration was 2.24 mg/ml, and the specific activity was8.74 x 106 U/mg of protein. Purity was >99%, and endotoxincontent was <0.0027 ng/mg of protein (Limulus test). TNFwas dissolved in sodium-phosphate buffer (50 mM) toachieve final concentrations of 2 x 105 U/ml. The dose usedof 4 x 105 U of TNF was selected to investigate the acuteeffects of TNF on clearance function independent of hemo-dynamic alterations. In pilot studies, this dose was shown toinduce inflammatory reactions, such as significant histaminerelease and eicosanoid generation, a marked activation ofgranulocytes with elastase release, and a moderate increasein pulmonary vascular resistance and microvascular perme-ability but without significant changes in systemic arterialpressure or oxygen uptake.Endotoxin from E. coli 0111 was kindly donated by R.

Urbaschek (Department of Immunology and Serology, Insti-tute of Medical Microbiology, Faculty of Clinical MedicineMannheim, University of Heidelberg). Lipopolysaccharide(LPS) was infused in a nonlethal dose, which did not inducesevere organ damage (e.g., tissue necrosis in the gastroin-testinal tract), as verified by light-microscope studies inprevious experiments during the observation period.

Bacterial inoculum. E. coli, an encapsulated, serum-resis-tant, nonhemolytic strain, freshly isolated from blood cul-ture of a septicemic patient, was cultivated on blood agarplates. The grown colonies were scraped from the plates,carefully homogenized by vortexing in tryptic soy broth,serially diluted, and adjusted to a density of 1.3 x 108CFU/ml. They were stored in a freezer until use. Theamount of exogenous E. coli used was based on pilotexperiments, investigating the clearance and organ distribu-tion of different doses of E. coli (1010, 108, 106). For thisstudy, a dose was chosen which showed an easily reproduc-ible elimination kinetic but did not induce severe hemody-namic changes influencing clearance function by tissue hy-poperfusion. The applied dose of 1.3 x 108 E. coli wascompletely eliminated during a time period of between 60and 90 min in control animals and allowed registration ofslight decreases as well as slight increases in eliminationkinetics during the observation period. At higher doses,delayed clearance might be due to the measured depressionof hemodynamics with increased vascular resistance, whichwas not the object of the present study. Shock-inducedimmune suppression has been reported in previous studies(1, 16). Experiments with lower doses, however, demon-strated such a rapid bacterial elimination (within 10 to 15min) that possible minor increases in clearance functionwere not detectable.

Statistical analysis. Data are presented as the arithmeticmean ± standard error. The logarithm of the quantitativebacterial count was used for statistical comparison. Differ-ences between groups were tested by one-way analysis ofvariance and subsequently verified by multiple-range test(Scheffe). Significance was accepted at P of <0.05.

This study was approved by the Animal Subject Protec-tion Committee of the University of Giessen. The care andhandling of animals were in accordance with the NationalInstitutes of Health guidelines.

INFECT. IMMUN.

on May 14, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

EFFECT OF ENDOTOXIN AND TNF ON BACTERIAL CLEARANCE 3145

TABLE 1. Counts of viable bacteria in organ homogenates taken180 min after injection of E. coli in the different groups

Loglo CFU of E. coli/g of tissue (mean + SEM)Organ

Controls TNF Endotoxin 1 Endotoxin 2

Liver 4.02 + 0.09 4.08 ± 0.04 4.47 ± 0.12a 5.51 ± 0.11bSpleen 3.86 ± 0.07 2.71 ± 0.14b 3.94 ± 0.13 5.34 ± 0.18bLung 1.71 ± 0.20 4.00 ± 0.06b 3.24 ± 0.10b 4.54 t 0.09bKidney 1.13 ± 0.28 1.90 ± 0.64 2.89 ± 0.09b 4.66 ± 0.07b

a P < 0.05 versus all other groups.b p < 0.001 versus all other groups.

0 30 60 90 120 150 180Time (min)

FIG. 1. Time course of bacterial elimination from blood afterinjection of E. coli (1.3 x 10' CFU) in the different groups. Meancounts of CFU were plotted semilogarithmically against time inminutes. The rapid decrease of bacteria within the first 10 min seenin all groups was followed by a significantly delayed blood clear-ance, with numbers of E. coli persisting up to 180 min in theendotoxemic animals, whereas cultures were sterile from 90 and 120min in the controls and TNF-challenged rabbits, respectively.

RESULTSHemodynamic and metabolic parameters. Mean measure-

ments of arterial pressure and oxygen uptake as well asblood gases (PaO2, PaCO2, SaO2, pH) did not differ signifi-cantly from baseline measurements throughout the studyperiod in the control and TNF-treated animals. In the rabbitsinfused with endotoxin for 4 h, however, arterial pressuredecreased progressively from 83 + 4 mm Hg (baseline) to 58+ 7 mm Hg (1 mm Hg = 133.322 Pa) at the end ofobservation (180 min after bacterial injection, P < 0.01versus controls; by one-way analysis of variance), whereasno major changes were evident in those infused with endo-toxin for 1 h. This pressure drop was paralleled by adecrease in pH from 7.48 + 0.01 to 7.26 + 0.05 comparedwith the other groups in which pH remained within thenormal range. Likewise, the impaired hemodynamics in theendotoxin 2 group could be correlated with the reduction ofoxygen uptake to 64% + 7% of the baseline value. Minimalvalues of oxygen uptake were 87% + 4%, achieved at theend of observation in the animals infused with endotoxin for1 h. There was a continuous increase in lactate concentra-tions (normal values in plasma, 5.7 to 22 mg/dl) in all groups;this increase was most pronounced in the endotoxin 2 group,with a more-than-fivefold increase of the baseline value from13.6 + 3.6 to 81.3 + 16.2 mg/dl. Minimal changes inhemoglobin and hematocrit, which were comparable in allgroups, ensued from dilution effects due to blood samplingwith isovolemic substitution with saline.

Microbiological cultures from blood samples taken beforeE. coli injection were sterile in all groups. The eliminationkinetics of bacteria from the circulating blood (Fig. 1)differed significantly between endotoxin-infused rabbits onthe one hand and TNF-treated animals and controls on theother hand. The rapid decrease of bacteria within the first 10min, which was seen in all groups, was followed by asignificantly delayed clearance with persisting numbers of E.coli until the end of observation in the endotoxemic animals.In contrast to this, blood cultures were sterile from 90 and120 min in the control and TNF-treated groups, respectively.

In comparison with the results in the controls, the delayedclearance in the rabbits challenged with endotoxin wasaccompanied by significantly higher numbers of E. coli in theliver, spleen, lung, and kidney (Table 1) and a partiallydifferent tissue distribution pattern of viable bacteria (Fig.2). Tissue cultures of the control group showed the highestbacterial numbers in the liver (57% + 6.7% of total CFU,counted in organ cultures) and in the spleen (43% ± 6.6%),whereas very low counts (<70 CFU/g) could be detected inthe lung (0.4% + 0.1%) and kidney (0.2% ± 0.1%). Incontrast to this, endotoxin (LPS) infusion for 4 h resulted insignificantly higher amounts of bacteria (100- to 1,000-fold ofcontrol values) in all four organs (P < 0.001 versus all othergroups) associated with significantly higher percentages oftotal bacteria detected in the lung (5% + 1.5%; P < 0.001,endotoxin 2 group versus control) and kidney (6% + 1.6%; P< 0.001, endotoxin 2 group versus control). Significantdifferences in the organ distribution pattern between bothendotoxemic groups were evident, with predominantamounts of E. coli in the liver (72% _ 4.5%; P < 0.001versus endotoxin 2 and TNF group) and minor numbers inthe spleen (2.5% ± 0.8%; P < 0.001 versus endotoxin 2 andTNF group) in the animals infused for 1 h, whereas highercounts in the kidney were detected in those infused for 4 h (P< 0.001 versus all other groups). TNF challenge produceddistinct alterations in absolute counts and especially in therelative tissue distribution pattern of bacteria, with markedlyincreased numbers of bacteria in the lung (44% ± 2.6%; P <

100- %

90-

80-

70-

60-

50-

40-

30-

20-

It -_*1u

O-

. .... *

......

...... 1

Liver 11iM Spleen

Ill 71 Lung IV EJ Kidney

Li11flL...~~~~~~I 1.fl........ I I U~~~~~~~~~. . ...I I -+11II IV 111II IV 111II IV 11111 IV

CONTROLS TNF ENDOTOXIN 1 ENDOTOXIN 2

FIG. 2. Comparative evaluation of organ distribution pattern ofviable CFU of E. coli expressed as percentage of total detectableCFU in tissue cultures. Relative counts varied significantly fromcontrols in endotoxin and TNF groups, with significantly highervalues in the lung and kidney and lower values in the spleen. *, P <0.05; **, P < 0.001 (by one-way analysis of variance).

VOL. 61, 1993

on May 14, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

3146 KOCH ET AL.

0.001) and very low counts in the spleen (2.5% ± 0.6%; P <0.001).

In additional pilot experiments, tests were performed todetermine whether or not the LPS and TNF treatment itselfleads to invasion of E. coli, particularly from the gut into theblood and other organs during the observation period. Theseexperiments revealed that bacterial translocation did nottake place during the experimental procedure in the absenceof injected E. coli in our model. Sterile blood and organcultures were found after TNF treatment (n = 3) or endo-toxin infusion for either 1 (n = 3) or 4 (n = 3) h withoutsubsequent E. coli injection. Consequently, the results rep-resent the decreasing numbers of circulating E. coli resultingexclusively from clearance function and not falsified mea-surements of incalculable alterations due to translocatedendogenous E. coli.

DISCUSSION

Severe injury predisposes the host to an increased suscep-tibility to infection, which often leads to acute respiratorydistress syndrome or multiple-organ failure (5). In view oftherapeutic strategies, several attempts have been made todelineate the underlying mechanisms of this complex rela-tionship. A link between shock and development of bacte-remia and endotoxemia was first suggested by Zweifach etal. (41) and Fine et al. (15). The role of shock-inducedimmune suppression was emphasized in studies reportingthat animals subjected to hemorrhagic shock cannot sustaina normal inflammatory response (1) and have impairedperitoneal clearance of bacteria (16). Clinical evidence indi-cating that the gut is a reservoir for bacteria causing systemicinfections in critically ill patients has accumulated. Loss ofintestinal bacterial barrier functions, with invasion of bacte-ria into the blood and translocation into other organs,appears to occur in certain clinical situations, resulting insystemic sepsis. Bacterial translocation in animal models (3,12, 21, 25) and increased intestinal permeability to lactuloseor mannitol in humans (8, 31) have been used as experimen-tal markers of such gut barrier dysfunction occurring afterhemorrhagic shock, endotoxin challenge, or thermal injury.Failure to eliminate translocated bacteria due to host immu-nosuppression may contribute to the persistence of entericbacteria in mesenteric lymph nodes and other organs (22, 24,32). It was previously reported that nonlethal doses ofendotoxin appear to promote bacterial translocation fromthe gut primarily by injuring the gut mucosa in mice (9, 10).The potential causal association between endotoxin-inducedintestinal injury and endotoxin-induced alterations in xan-thine oxidase activities and bacterial translocation (14) is notfully clarified. In addition to questions on the effects ofendotoxin on intestinal barrier function, another importantquestion is whether endotoxin directly or indirectly influ-ences phagocytosis and lysis capacity of the RES, thusenhancing bacterial growth. It appears that endotoxin pro-duces its toxic effects in vivo by stimulating host cells,especially macrophages, to release various proinflammatorymediators that act as secondary messengers (20, 27). TNF issuch a factor that has been found to play a pivotal role in thedevelopment of shock and tissue injury during septicemia(39, 40). TNF has been documented to induce shock, tissueinjury, and necrosis of tumor cells as well as tissue necrosisin the gastrointestinal tract (37). Furthermore, it is knownthat TNF causes the release of other mediators such asinterleukin 1, prostaglandins (2, 29), platelet-activating fac-tor (6), and collagenases (7) and directly activates neutro-

phils and promotes adhesion between polymorphonuclearleukocytes and endothelial cells (17, 23).

Since endotoxemia is relatively common after severeinjury and is associated with conditions leading to multiple-organ failure, the aim of this study was to evaluate thepotential influence of endotoxin on bacterial clearance and todetermine the role ofTNF in mediating septic complications.In contrast to others (3, 9-14), the present report was notintended to investigate possible enhanced translocation fromthe gut but whether systemic and organ clearance of system-ically applied E. coli is impaired after infusion of endotoxinfor different time periods and after TNF injection. Theintravenous application of exogenous E. coli was chosen asa correlate of bacterial invasion from various compartments,e.g., the gut, the urogenital tract, wounds, or implantedcatheters, into the circulating blood. To enable quantifica-tion of the clearance process, a defined number of E. coliwas injected intravenously instead of inducing the invasionof uncontrollable numbers of bacteria from the gut into thecirculation. To exclude possible miscalculation resultingfrom incalculable numbers of bacteria being translocatedfrom the gut, additional experiments were performed todetermine whether or not the LPS and TNF treatment itselfleads to invasion of endogenous E. coli from the gut into theblood and to translocation into tissues during the observa-tion period. These experiments revealed that bacterial trans-location did not take place, as evidenced by sterile culturesin our model. Hence, enhanced translocation of intestinal E.coli, which might mask the effects on bacterial clearancefunction and tissue distribution, could be excluded. Thus,the differences in numbers of detected CFUs could beaccounted for by alterations in clearance function due to theapplied injury.The present results demonstrate distinct alterations in

bacterial elimination from the blood in the endotoxin-chal-lenged groups and significantly different organ distributionpatterns of bacteria in the endotoxin- and TNF-treatedanimals compared with those of saline-injected control rab-bits. The elimination kinetics (Fig. 1) of injected E. colisuggest a biphasic time course, which is characterized by arapid phase described by an exponential-like decrease ofCFUs in the first 15 min, followed by a subsequent, slowerphase of bacterial clearance. The second phase of bacterialclearance was significantly prolonged in both endotoxemicgroups independent of the duration of the preceding endo-toxin infusion, whereas the initial phase of rapid bacterialelimination was less altered in these groups. Whether ablockade of the RES by endotoxin, with consequentiallyreduced phagocytosis and lysis capacity, is one potentialmechanism responsible for the failure to completely removebacteria from the circulation remains to be clarified. Thedelayed bacterial clearance associated with significantlyhigher numbers of viable bacteria in organs was morepronounced after 4 h of endotoxin infusion than in the groupinfused for 1 h. In the latter group, it seems that thepredominant amount of injected bacteria is filtered off andpotentially phagocytized in the liver (72%), whereas after 4 hof infusion, a different distribution pattern with highercounts in the spleen, lung, and kidney was found. One canonly speculate as to the underlying mechanisms leading tothese findings. Apart from impaired phagocytotic and lyticactivity of various cell lines, such as neutrophils or circulat-ing or tissue macrophages, a saturation of the phagocyticcells by ingested particles (e.g., endotoxin, coagulationproducts, debris), as well as a potential deficit of humoralphagocytosis-supporting factors (as discussed in previous

INFECT. IMMUN.

on May 14, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

EFFECT OF ENDOTOXIN AND TNF ON BACTERIAL CLEARANCE 3147

studies [24]), may contribute to the reported results. Similarfindings, demonstrating an augmented extrahepatic, espe-cially pulmonary, localization of systemically applied micro-particles at times of hepatic clearance depression, wereobserved in other experimental studies in rats (34, 35). Inattempting to analyze the mechanism leading to alteredorgan distribution patterns, regional blood flow was investi-gated. However, no correlation between decreased bloodflow and organ distribution could be found (22). In a primatemodel, it was shown that the baboon responds to a nearlylethal dose of endotoxin, which severely depresses arterialblood pressure, without a decrease in mesenteric blood flow(38). Thus, if the RES was to be compromised by thehypotensive effects of endotoxemia, it does not appear to becaused by a decrease in organ blood flow. Hence, it isunlikely that altered organ distribution in the endotoxin 2group is due to the decreased hemodynamics observed at theend of the study period. In our model, invasive measure-ments of regional blood flow were avoided because ofpotential infections and mediator release, which can presum-ably influence the microbiological results in an unpredictablemanner.To elucidate a possible modifying effect of the cytokine

TNF on bacterial clearance independent of hemodynamicalterations, human recombinant TNF was injected 60 minprior to bacterial application in a dose which did not produceshock symptoms. Since a great variation in the levels ofendogenous TNF production and responses to exogenousTNF administration among different species can be assumed(4, 18, 19), a bolus dose of 4 x 105 U of TNF, like that ofprior experiments with rabbits in our laboratory, was used.This dose produced moderate increases in pulmonary resis-tance and microvascular permeability, as well as in hista-mine liberation, and a marked activation of granulocytes butwithout hemodynamic effects (unpublished data). In contrastto the endotoxin infusion, the TNF treatment did not essen-tially influence the elimination kinetics, possibly because ofTNF-induced neutrophil activation (23) and enhancedphagocytosis (28), but did distinctly alter the organ distribu-tion of viable bacteria. Deviating from the pattern in thecontrol and endotoxemic animals, high tissue counts equal toa relative organ uptake of 44% were discovered in the lungand low counts (2.5%) were found in the spleen. The highnumbers of E. coli found in the lung may be due to enhancedleukocyte sticking in the lung, reflecting a state of hyper-phagocytotic activity. Since leukocyte accumulation andadherence is a well-known cytokine-induced phenomenon,an increase of phagocytotic cells in the lung, per se, mayaccount for the elevated bacterial counts in the TNF-treatedanimals compared with those of the endotoxin-treated andcontrol animals. No attempts were made in the present studyto differentiate between intra- and extracellular viable bac-teria, but such information might be of value as an indicatorfor local phagocytotic activity. Future studies correlatingbacterial growth with histological data and quantitativeevaluations of phagocytotic cells are required to clarify thiscomplex pathophysiological relationship.Summarizing the reported results, endotoxin was shown

to induce impaired bacterial clearance from the blood thatwas associated with increased bacterial counts in tissues,thus reflecting reduced phagocytosis and lysis capacity ofthe RES. These findings point towards a weaker resistanceagainst bacterial colonization and translocation from innerand outer body surfaces and may account for the highincidence of sepsis in severely injured patients. The influ-ence of TNF, which is known to modulate immune response,

resulted in a distinctly different organ distribution pattern,with bacterial growth predominating in the lung, whileelimination kinetics were unaltered. Since endogenous TNFmay affect elimination kinetics not seen with bolus TNFadministration, the question of whether TNF also exerts amediating role on the bacterial blood clearance remains to beclarified in additional studies dealing with TNF antibodies.The current results warrant further investigations to eluci-date the relative role of cytokines and the interrelationshipwith other mediators (platelet-activating factor, eicosanoids,free radicals) in the pathogenesis of endotoxin-induced RESimpairment.

ACKNOWLEDGMENTSWe thank A. Weber and U. Gerhardt for excellent technical

assistance and T. Wieth for proofreading the manuscript.

REFERENCES1. Abraham, E., and Y.-H. Chang. 1987. Effects of hemorrhage on

inflammatory response. Arch. Surg. 119:1154-1157.2. Bachwich, P. R., S. W. Chensue, J. W. Larrick, and S. L.

Kunkel. 1986. Tumor necrosis factor stimulates interleukin 1and prostaglandin E2 production in resting macrophages. Bio-chem. Biophys. Res. Commun. 136:94-101.

3. Baker, J. W., E. A. Deitch, R. D. Berg, and R. D. Specian. 1988.Hemorrhagic shock induces bacterial translocation from thegut. J. Trauma 28:896-904.

4. Beutler, B. A., I. W. Milsark, and A. Cerami. 1985. Cachectin/tumor necrosis factor: production, distribution, and metabolicfate in vivo. J. Immunol. 136:3972-3977.

5. Carrico, C. J., J. L. Mealdns, J. C. Marchall, D. Fry, and B. V.Maier. 1986. Multiple organ failure syndrome. Arch. Surg.121:196-201.

6. Chang, S. W., C. 0. Feddersen, P. M. Henson, and N. F.Voelkel. 1987. Platelet-activating factor mediates hemodynamicchanges and lung injury in endotoxin treated rats. J. Clin.Invest. 79:1498-1509.

7. Dayer, J. M., B. Beutler, and A. Cerami. 1985. Cachectin/tumornecrosis factor stimulates collagenase and prostaglandin E2production by human synovial cells and dermal fibroblasts. J.Exp. Med. 162:2163-2168.

8. Deitch, E. A. 1990. Intestinal permeability is increased in burnpatients shortly after injury. Surgery 107:411-416.

9. Deitch, E. A., R. D. Berg, and R. D. Specian. 1987. Endotoxinpromotes the translocation of bacteria from the gut. Arch. Surg.122:185-190.

10. Deitch, E. A., L. Ma, W. J. Ma, M. B. Grisham, D. N. Granger,R. D. Specian, and R. D. Berg. 1989. Inhibition of endotoxin-induced bacterial translocation in mice. J. Clin. Invest. 84:36-42.

11. Deitch, E. A., and R. McIntyre Bridges. 1987. Effect of stressand trauma on bacterial translocation from the gut. J. Surg. Res.42:536-542.

12. Deitch, E. A., J. Morrison, R. D. Berg, and R. D. Specian. 1990.Effect of hemorrhagic shock on bacterial translocation, intesti-nal morphology, and intestinal permeability in conventional andantibiotic-decontaminated rats. Crit. Care Med. 18:529-536.

13. Deitch, E. A., M. Taylor, M. Grisham, L. Ma, W. Bridges, andR. Berg. 1989. Endotoxin induces bacterial translocation andincreases xanthine oxidase activity. J. Trauma 29:1679-1683.

14. Deitch, E. A., M. Taylor, M. Grisham, L. Ma, W. Bridges, andR. Berg. 1989. Endotoxin induces bacterial translocation andincreases xanthine oxidase activity. J. Trauma 29:1679-1683.

15. Fine, J., S. Ruthenberg, and F. B. Schweinburg. 1959. The roleof the reticuloendothelial system in hemorrhagic shock. J. Exp.Med. 110:547-569.

16. Fink, M. P., M. Gardiner, and T. J. MacVittie. 1985. Sublethalhemorrhage impairs acute peritoneal response in rats. J. Trauma25:234-237.

17. Gamble, J. R., J. M. Harlan, S. J. Klebanoff, and M. A. Vadas.1985. Stimulation of the adherence of neutrophils to umbilical

VOL. 61, 1993

on May 14, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

3148 KOCH ET AL.

vein endothelium by human recombinant tumor necrosis factor.Proc. Natl. Acad. Sci. USA 82:8667-8671.

18. Girardin, E., G. E. Grau, J. M. Dayer, P. Roux-Lombard, andP. H. Lambert. 1988. Tumor necrosis factor and interleukin-1 inthe serum of children with severe infectious purpura. N. Engl. J.Med. 319:397-400.

19. Haranaka, K., E. A. Carswell, B. D. Williamson, J. S. Prender-gast, N. Satomi, and L. J. Old. 1986. Purification, characteriza-tion, and antitumor activity of nonrecombinant mouse tumornecrosis factor. Proc. Natl. Acad. Sci. USA 83:3949-3953.

20. Hesse, D. G., K. J. Tracey, Y. Fong, K. R. Manogue, M. A.Palladino, Jr., A. Cerami, G. T. Shires, and S. F. Lowry. 1988.Cytokine appearance in human endotoxemia and primate bac-teremia. Surg. Gynecol. Obstet. 166:147-153.

21. Jones, W. G., A. E. Barber, J. P. Minei, T. J. Fahey, G. T.Shires, and G. T. Shires. 1990. Differential pathophysiology ofbacterial translocation after thermal injury and sepsis. Ann.Surg. 241:24-30.

22. Kaplan, J. E., and T. M. Saba. 1976. Humoral deficiency andreticuloendothelial depression after traumatic shock. Am. J.Physiol. 230:7-14.

23. Klebanoff, S. J., M. A. Vadas, J. M. Harlan, L. H. Sparks, J. R.Gamble, J. M. Agosti, and A. M. Waltersdorph. 1986. Stimula-tion of neutrophils by tumor necrosis factor. J. Immunol.136:4220-4225.

24. Loegering, D. J. 1977. Humoral factor depletion and reticuloen-dothelial depression during hemorrhagic shock. Am. J. Physiol.232:283-287.

25. Morris, S. E., N. Navaratnam, and D. N. Herndon. 1990. Acomparison of effects of thermal injury and smoke inhalation onbacterial translocation. J. Trauma 30:639-643.

26. Morrison, D. C., and J. L. Ryan. 1979. Bacterial endotoxins andhost immune response. Adv. Immunol. 28:293-450.

27. Morrison, D. C., and J. L. Ryan. 1987. Endotoxins and diseasemechanisms. Annu. Rev. Med. 38:417-432.

28. Moxey-Mims, M. M., H. H. Simms, M. M. Frank, E. Y. Lin, andT. A. Gaither. 1991. The effects of IL-1, IL-2, and tumornecrosis factor on polymorphonuclear leukocyte Fcy receptor-mediated phagocytosis. J. Immunol. 147:1823-1830.

29. Nawroth, P. P., I. Bank, D. Handley, J. Cassimeris, L. Chess,and D. Stern. 1986. Tumor necrosis factor/cachectin interacts

with endothelial cell receptors to induce release of interleukin 1.J. Exp. Med. 163:1363-1375.

30. Neuhof, H. 1981. Monitoring of total oxygen consumption forcontrol of patients during extracorporeal circulation, p. 266-270. In H. P. Kimmich (ed.), Monitoring of vital parametersduring extracorporeal circulation. Karger, Basel.

31. O'Dwyer, S. T., H. R. Michie, T. R. Ziegler, A. Revhaug, R. J.Smith, and W. Wilmore. 1988. A single dose of endotoxinincreases permeability in healthy humans. Arch. Surg. 123:1459-1464.

32. Penn, R. L., R. D. Maca, and R. D. Berg. 1985. Increasedtranslocation of bacteria from the gastrointestinal tracts oftumor-bearing mice. Infect. Immun. 47:793-798.

33. Rush, B. F., A. J. Sori, F. M. Thomas, S. Smith, J. J. Flanagan,and G. W. Machiedo. 1988. Endotoxemia and bacteremia duringhemorrhagic shock. Ann. Surg. 207:549-554.

34. Saba, T. M. 1972. Effect of surgical trauma on the clearance andlocalization of blood-borne particulate matter. Surgery 71:675-685.

35. Saba, T. M., and T. G. Antikatzides. 1979. Heparin inducedalterations in clearance and distribution of blood-borne micro-particles following operative trauma. Ann. Surg. 189:426-432.

36. Sandys, G. H. 1960. A new method of preventing swarming ofProteus sp. with a description of a new medium suitable for usein routine laboratory practice. J. Med. Lab. Technol. 17:224-233.

37. Sun, X. M., and W. Hsueh. 1988. Bowel necrosis induced bytumor necrosis factor in rats is mediated by platelet-activatingfactor. J. Clin. Invest. 81:1328-1331.

38. Swan, K. G., and D. G. Reynolds. 1972. Blood flow to the liverand spleen during endotoxin shock in the baboon. Surgery72:388-394.

39. Tracey, K. J. 1991. Tumor necrosis factor (cachectin) in thebiology of septic shock syndrome. Circ. Shock 35:123-128.

40. Tracey, K. J., and S. F. Lowry. 1990. The role of cytokinemediators in septic shock. Adv. Surg. 23:21-56.

41. Zweifach, B. W., B. Benacerraf, and L. Thomas. 1957. Therelationship between the vascular manifestations of shock pro-duced by endotoxin, trauma, and hemorrhage. II. The possiblerole of the reticuloendothelial system in resistance to each typeof shock. J. Exp. Med. 106:403-414.

INFECT. IMMUN.

on May 14, 2020 by guest

http://iai.asm.org/

Dow

nloaded from