INFECTION AND IMMUNITY, Aug. 2002, p. 44414446 Vol. 70, No. 80019-9567/02/$04.000 DOI: 10.1128/IAI.70.8.44414446.2002Copyright 2002, American Society for Microbiology. All Rights Reserved.

Interleukin-10 Controls the Onset of Irreversible Septic ShockSamir Q. Latifi,1 Mary Ann ORiordan,1 and Alan D. Levine2*

Division of Pediatric Pharmacology and Critical Care Medicine, Department of Pediatrics,1 and Department of Medicine, Pathology,and Pharmacology,2 Case Western Reserve University School of Medicine, Cleveland, Ohio 44106-4952

Received 3 April 2002/Accepted 29 April 2002

Lethality from sepsis is believed to be mediated by a proinflammatory cytokine cascade, yet blocking theproinflammatory cytokines tumor necrosis factor alpha (TNF-) and interleukin-1 (IL-1) fails to preventmortality in human disease and a mouse model of sepsis induced by cecal ligation and puncture (CLP). Therole of the antiinflammatory cytokine IL-10 in the CLP model of sepsis is unclear, with either protective orharmful effects demonstrated, depending upon the time of intervention. We therefore hypothesize that IL-10functions as a temporal regulator of the transition from early reversible sepsis to the late phase of irreversibleshock. Transition from reversible sepsis to irreversible shock in the CLP model was defined as the time whenremoval of the necrotic cecum by rescue surgery is no longer effective. We subjected IL-10-deficient (IL-10/)and wild-type (IL-10/) mice to CLP and monitored the progression of sepsis, the onset of irreversible shock,and mortality. Onset of lethality in IL-10/ mice occurred significantly earlier than in IL-10/ mice and wasassociated with 15-fold-higher serum levels of TNF- and IL-6. Consistent with these findings, the efficacy ofrescue surgery after lethal CLP is lost 10 h earlier in IL-10/ mice than in IL-10/ mice. Treatment withrecombinant human IL-10 5 h after CLP significantly improved survival and lengthened the therapeuticwindow for rescue surgery in both strains of mice. These results demonstrate that IL-10 controls the onset ofirreversible septic shock after CLP.

Sepsis is a major cause of morbidity and mortality in ourhospitals (2). A severe and uncontrolled systemic inflammatoryresponse triggered by an invading microbe may lead to evolv-ing multiorgan dysfunction (7). Occasionally, despite appropri-ate treatment and support of the septic patient, death may stillensue if irreversible damage occurs to vital organs. The onsetof this irreversible shock is not easily defined clinically, nor arethe mechanisms regulating this transition known. Cytokinesmay be important mediators in the development of this lethalmultiorgan damage. For example, inflammatory cytokines,such as tumor necrosis factor alpha (TNF-), interleukin-1(IL-1), and IL-6, are elevated in the sera of both adult andpediatric septic patients (6, 9, 20, 29) and affect symptoms suchas hypotension, fever, and the production of acute-phase pro-teins (21). However, the contribution of these proinflammatorycytokines in directly mediating mortality from sepsis is notclear, since therapies utilizing neutralizing antibodies or solu-ble receptor antagonists against TNF- and IL-1 fail to show asignificant benefit in outcome for patients with sepsis (5, 13).Possible explanations for the clinical failure of these cytokine-targeted therapies may be that the antibodies or antagonistswere administered too late or that other unopposed inflamma-tory mediators continue to support the disease process.

In contrast to humans, the use of anticytokine therapiesagainst TNF- (4) and IL-1 (22) in a murine model of endo-toxemia leads to dramatic improvement in animal mortality.When lipopolysaccharide (LPS), the endotoxin from gram-negative bacteria, is injected into susceptible animals a rapidsystemic inflammatory cascade is initiated in the absence of

bacteremia. The lack of ongoing injury after a single adminis-tration of endotoxin and the absence of bacteremia in thismodel highlight its artificial nature and may explain why manyclinical trials, based on observations with this model, havefailed. The kinetics and magnitude of cytokine production inLPS-induced sepsis are also dramatically different from thatseen in a more clinically relevant model of peritonitis withbacteremia induced by cecal ligation and puncture (CLP) (25),which closely mimics the human disease of septic shock (14). Insepsis induced by CLP, neutralizing antibodies to TNF- arenot protective against mortality (10), and C3H/HeJ mice thatare resistant to the lethal effects of LPS are still susceptible tomortality from CLP (19). These differences between the LPSand CLP models suggest that the immunology of sepsis is farmore complex than the cascade of events activated by endo-toxin and that the exact role of proinflammatory cytokines asmediators of mortality is unclear.

Besides stimulating the synthesis of proinflammatory cyto-kines, the septic response in patients and animal models alsoresults in the production of antiinflammatory mediators, suchas IL-10 (8, 11, 30, 31). IL-10 is a 35-kDa homodimeric protein,produced primarily by monocytes, T cells, and B cells, thatinhibits the production of proinflammatory cytokines in vitro(16). IL-10 is completely protective in the LPS model of sepsis(3, 15), but in the CLP model the administration of neutraliz-ing antibodies to IL-10 at the time of CLP only partially exac-erbates mortality (28, 31). However, inhibition of IL-10 12 hafter CLP actually improves survival (28). Thus, in this modelIL-10 can be either protective or harmful depending on thetime of intervention. We therefore postulate that IL-10 regu-lates the transition from reversible sepsis to irreversible shock.In the CLP model we define this transition as the time pointwhen removal of the necrotic cecum by rescue surgery is nolonger effective. Here we report the ability of IL-10 to regulate

* Corresponding author. Mailing address: Department of Medicine,Case Western Reserve University School of Medicine, 10900 EuclidAve., Cleveland, OH 44106-4952. Phone: (216) 368-1668. Fax: (216)368-1674. E-mail: adl4@po.cwru.edu.

4441

the progression of sepsis and control the onset of irreversibleshock and mortality in the CLP model.

MATERIALS AND METHODS

Mice. Inbred C57BL/6J mice deficient for the IL-10 gene (IL-10/) andwild-type littermates (IL-10/) were used for experimentation between the agesof 8 to 12 weeks. The mice were bred and maintained in microisolator cages inour specific-pathogen-free barrier facility under a 12-h light-dark cycle. WhileIL-10/ mice are known to develop colitis, in our facility no signs or symptomsof colitis are observed until 14 weeks of age. All experiments described wereperformed in adherence to the National Institutes of Health guidelines on theuse of experimental animals, and approval was obtained from the InstitutionalAnimal Care and Use Committee of Case Western Reserve University.

Genotyping of IL-10/ mice. The genotype of each mouse was determined byPCR analysis from tail clip DNA (12). Briefly, tail clips 0.25 cm in length weredigested with proteinase K (Gibco-BRL, Gaithersburg, Md.) in tail buffer (50mM Tris buffer, pH 8.0; 100 mM EDTA; 100 mM NaCl; 1% sodium dodecylsulfate) overnight. The following day, digested tails were extracted twice withphenol (Gibco-BRL), followed by an extraction once with 24:1 chloroform-isoamyl alcohol (Sigma, St. Louis, Mo.). DNA was precipitated with 100%isopropanol, washed in 70% ethanol, air dried, and dissolved in 100 l of TEbuffer (10 mM Tris, 1 mM EDTA; pH 8.0). A total of 1 l of DNA was addedto a 25-l PCR containing 1 Taq extender buffer (Stratagene, La Jolla, Calif.),0.25 mM deoxynucleoside triphosphates (Gibco-BRL), 2 U of Taq polymerase(Boehringer Mannheim, Indianapolis, Ind.), and 2 U of Taq Extender (Strat-agene). Then, 1 l of a three-primer cocktail (1.6 g/ml) was added (primer 1,5-GCCTTCAGTATAAAAGGGGGAC-3; primer 2, 5-CAGTTCATTCAGGGCACCGG-3; primer 3, 5-CTTCAAACCCCCAAACCTCTG-3). PCR am-plification conditions were 94C for 5 min, 94C for 1 min, 55C for 1 min, and72C for 1 min for 40 cycles, followed by 72C for 5 min. The products wereanalyzed on a 2.0% agarose gel in 0.5 Tris-borate-EDTA buffer. Samples wereloaded in buffer containing bromophenol blue and xylene cyanol. The gel wasstained with 1 g of ethidium bromide/ml for 30 min and then destained indistilled water for an additional 30 min. The product size for the wild-type anddisrupted IL-10 genes were 700 and 600 bp, respectively.

Cecal ligation and puncture. Anesthesia in the mice was induced with anintraperitoneal injection of 2.5% tribromoethanol at 0.02 ml/g (23). The abdo-men was opened, the cecum identified and ligated with a 3-O silk suture at itsbase below the ileocecal junction. Patency of the bowel lumen from the ileum tothe colon was maintained. The antimesenteric border of the cecum was punc-tured once with either a 25- or 18-gauge needle, and the cecum was gentlysqueezed to extrude a small amount of stool. The cecum was returned to theabdomen, which was closed in two layers with 5-O silk sutures. After surgerymice were resuscitated with 0.5 ml of normal saline given by subcutaneousinjection. Mice received 25 mg of the broad-spectrum antibiotic imipenem(Merck, West Point, Pa.)/kg in 0.5 ml of normal saline twice daily by subcuta-neous injection for 3 days. Mice were observed for a minimum of 120 h afterCLP, after which a live animal was considered a long-term survivor. For miceundergoing sham surgery, the procedure for CLP was followed as describedabove except the cecum was not ligated and punctured.

Cecal removal (CR) surgery. After CLP some mice underwent a second rescuesurgery, wherein the necrotic cecum was excised to remove the infecting nidus.Before the septic mice were anesthetized for their second laparotomy, they werefurther fluid resuscitated by a subcutaneous injection of 0.5 ml of normal saline.A 30% lower dose of tribromoethanol was administered by intraperitoneal in-jection to ensure recovery from anesthesia by all animals. The abdomen of eachanimal was opened, and the necrotic cecum was excised with scissors, whileensuring that the silk tie at the base of the cecum remained intact to preventleakage of bowel contents. The peritoneal cavity was then lavaged with 1 ml ofnormal saline, and the abdomen was closed in two layers with 5-O silk sutures.After recovery from surgery, observation of the mice continued as before.

Blood sample collection for cytokine analysis. For cytokine analysis, blood wascollected either by tail vein bleed or cardiac puncture in the euthanized animal.No mouse was bled more than twice; only 50 to 100 l was taken each time tominimize iatrogenic blood loss. Serum was stored at 70C for later cytokineassay.

Cytokine ELISAs. The concentrations of TNF-, IL-6, and IL-10 in serumwere measured by enzyme-linked immunosorbant assay (ELISA) as recom-mended by the manufacturer (R&D Systems, Minneapolis, Minn.). The sensi-tivity of the IL-6 and the IL-10 ELISA kits was 15 pg/ml, and for the TNF- kitit was 23 pg/ml.

Treatment of mice with IL-10. In specific experiments, 1 g of recombinanthuman IL-10 (rhIL-10 [a gift from Schering-Plough Research Institute, Ken-ilworth, N.J.]) was administered after CLP by subcutaneous injection in 0.25 mlof normal saline, whereas control mice received vehicle alone. The total dailyfluid administered to the mice after CLP remained the same by an appropriatedecrease in the volume of the subsequent dose of antibiotic. The dose of IL-10used here has previously been shown to be protective in the LPS model of sepsis(15).

Data analysis. The overall survival for lethal and sublethal injury was de-scribed by using Kaplan-Meier plots. The difference between the survival curvesfor IL-10/ and IL-10/ mice was tested by using the Logrank test. Propor-tions alive at specific time points were compared by using Fishers exact test.Cytokine data were transformed to the log10 scale, and the results were plottedas means and standard deviations of the data. Two-way analysis of variance(ANOVA) was used to test for the main effects of time after CLP and group(IL-10/ versus IL-10/), as well as the interaction between time and group.Analyses for TNF- and IL 6 were performed separately. Time of death of theanimals used in these experiments was included in the model as a covariate. Allanalyses were done by using the SAS System (v8.1; SAS Institute, Carey, N.C.).

RESULTS

Endogenous IL-10 delays the onset of lethality induced byCLP and protects against mortality. To test our hypothesisthat IL-10 functions as a temporal regulator of the septic re-sponse, we compared the time of onset of mortality and thesubsequent kinetics of disease progression between IL-10-de-ficient mice on a C57BL/6J background (IL-10/) and theirwild-type littermates (IL-10/) after CLP with an 18-gaugeneedle (Fig. 1). Onset of lethality in IL-10/ mice occurred at10 h after CLP, but in IL-10/ mice it was delayed until 30 h,by which time the IL-10/ mice had a significant 67% greatermortality (Fishers exact test [FET], P 0.0001). The overallmortality in the IL-10/ mice was also significantly morerapid when we compared the survival curves between bothstrains of mice (Logrank test, P 0.0001). However, the mod-est 15% difference in long-term survival at 120 h after CLP

FIG. 1. In the absence of IL-10 mortality is more rapid after lethalCLP. IL-10/ (dashed line) and IL-10/ (solid line) C57BL/6J miceunderwent CLP with an 18-gauge needle (n 30 mice per strain). Thetime of death within 10-h intervals was recorded, and the data areexpressed as the cumulative percentage of mice still alive within eachinterval. The overall mortality in the IL-10/ mice was significantlymore rapid when the survival curves between both strains of mice werecompared (Logrank test, P 0.0001), but the mortality at 120 h wasstatistically indistinguishable between the two strains (FET, P 0.20).

4442 LATIFI ET AL. INFECT. IMMUN.

between IL-10/ and IL-10/ mice was not statistically sig-nificant (FET, P 0.20). These results indicate that the ab-sence of endogenous IL-10 leads to a more rapid onset ofdeath but does not affect the overall mortality.

Delay of disease onset by IL-10 is similarly observed after asublethal injury. The more rapid onset of mortality in theIL-10/ mice compared to IL-10/ animals after CLP withan 18-gauge needle may be due to the severity of the insult. Wetherefore evaluated a sublethal injury with a 25-gauge needleto puncture the cecum and then monitored survival (Fig. 2).The onset of mortality at 10 h after CLP in IL-10/ mice wasidentical between a lethal and sublethal injury. In IL-10/

mice the onset of mortality after sublethal injury was delayedfor 60 h, by which time there was a 35% significantly greatermortality in the IL-10/ mice (FET, P 0.01). Similar to thefindings from the more lethal CLP, overall survival at 120 h wasnot significantly different (FET, P 0.52). These results dem-onstrate that endogenous IL-10 has little influence on long-term survival but delays the onset of mortality.

Kinetics of proinflammatory cytokine production after CLP.An IL-10 deficiency has greater impact on mortality earlier inthe disease process, suggesting that IL-10 functions in the CLPmodel of sepsis to regulate the timeline of systemic inflamma-tion. We therefore predicted that the absence of IL-10 wouldalter the kinetics of production of the proinflammatory cyto-kines TNF- and IL-6 after CLP. IL-10/ mice producedsignificantly greater levels of TNF- and IL-6 in serum com-pared to IL-10/ animals at each time point from 5 to 20 h(ANOVA, P 0.0001) after CLP with an 18-gauge needle(Fig. 3). In comparison, IL-10/ and IL-10/ mice that un-derwent sham surgery had no detectable TNF- in their serumand only a small peak of IL-6, to 1 ng/ml, at 5 h afterlaparotomy (data not shown). There was no significant changefrom 5 to 20 h after CLP in the concentrations of either TNF-or IL-6 in the sera in both strains of mice (ANOVA, P 0.07).

The 15-fold-greater concentrations of TNF- and IL-6 mea-sured in the sera of IL-10/ mice after CLP may contribute tothe earlier onset of lethality.

IL-10 levels in serum are elevated after CLP. The lowerTNF- and IL-6 levels seen after CLP in IL-10/ mice com-pared to IL-10/ mice are not unexpected. Therefore, weinvestigated the kinetics of serum IL-10 production in IL-10/ mice after lethal CLP to determine whether it paralleledthe proinflammatory cytokine profile. Levels of IL-10 in serumwere maximal at 5 h after CLP and remained elevated until20 h after injury (Fig. 4), matching the kinetics of proinflam-matory cytokine production. As expected, IL-10/ mice didnot have any detectable IL-10.

Absence of IL-10 shortens the therapeutic window for rescuesurgery after lethal CLP. Our findings thus far indicate thatIL-10 regulates the onset of mortality, suggesting that IL-10may also modulate the transition from early reversible sepsis tolate irreversible shock. It was shown previously that the lethal-ity of CLP could be reversed by a second surgery to remove theligated necrotic cecum (1). Therefore, we defined the transi-tion from reversible sepsis to irreversible shock in the CLPmodel as the time when rescue surgery by removal of the

FIG. 2. Delayed onset of mortality in the presence of IL-10 aftersublethal CLP. IL-10/ (dashed line) and IL-10/ (solid line) miceunderwent CLP with a 25-gauge needle, and the time of death aftersurgery was recorded within 10-h intervals (n 20 mice per strain). Atthe onset of lethality in IL-10/ mice at 60 h there was a 35%significantly greater mortality in the IL-10/ mice (FET, P 0.01).However, there was no statistical difference in long-term survival be-tween IL-10/ and IL-10/ mice (FET, P 0.52).

FIG. 3. The absence of IL-10 led to a greater elevation in TNF-and IL-6 in serum during sepsis initiated by CLP. IL-10/ (F) andIL-10/ (E) mice that underwent lethal CLP had sera collected at theindicated times, and TNF- (A) and IL-6 (B) concentrations weredetermined by ELISA (n 8 mice per time point). The data are shownas the log transformation of the mean the standard deviation.

VOL. 70, 2002 IL-10 CONTROLS THE ONSET OF IRREVERSIBLE SEPSIS 4443

necrotic cecum is no longer effective. At the time of CR allmice exhibited similar signs of murine sepsis, such as ruffledfur, periorbital exudates, tremor, and lethargy. IL-10/ micecould be partially rescued by CR at 5 h after lethal CLP(18-gauge needle), with 60% surviving long term compared to3% after CLP alone (FET, P 0.001; Fig. 5). At 10 h CR wasineffective in IL-10/ mice, with the 20% survival being notsignificantly different to that after CLP alone (FET, P 0.15),whereas IL-10/ mice that underwent CR at 10 h still showed100% survival compared to the 17% survival with no interven-tion (FET, P 0.0001). However, by 20 h CR in IL-10/ micewas no longer efficacious (FET, P 1.0). These findings showthat in lethal sepsis induced by CLP the success of rescuesurgery is time dependent. Furthermore, the earlier failure ofCR in IL-10/ mice suggests that IL-10 regulates the transi-tion from reversible sepsis to irreversible shock.

Treatment with IL-10 improves survival and delays the on-set of mortality and irreversible shock. As IL-10 appears tomodulate the therapeutic window for rescue surgery, thentreatment with exogenous IL-10 should delay the onset ofirreversible shock. We therefore treated both IL-10/ andIL-10/ mice with 1 g of rhIL-10 after CLP to determinewhether the onset of lethality would be delayed and the ther-apeutic window for rescue surgery extended. The time chosenfor administration of the rhIL-10 was 5 h after CLP becausethis was the earliest time point at which we found maximalIL-10 levels in the sera of IL-10/ mice (Fig. 4). In bothstrains of mice after CLP with an 18-gauge needle, rhIL-10treatment delayed the onset of lethality compared to mice whoreceived no treatment (Fig. 6 and Table 1). Furthermore, forboth IL-10/ and IL-10/ mice treated with rhIL-10 5 hafter CLP overall survival was significantly improved comparedto that of untreated mice of the same strain (Logrank test, P 0.007). Consistent with previous reports (17, 24), treatment of

FIG. 4. The IL-10 concentrations in serum are elevated and main-tained from 5 to 20 h during sepsis initiated by CLP. Groups ofIL-10/ mice underwent lethal CLP and had sera collected at theindicated times (n 8 mice per time point). IL-10 concentrations inserum were determined by ELISA. The data are shown as the logtransformation of the mean the standard deviation.

FIG. 5. Rescue surgery with CR must occur earlier in IL-10/

mice to improve survival. Groups of IL-10/ (, n 10) and IL-10/(, n 10) mice underwent CR at 5, 10, 15, and 20 h after CLP, andthe survival at 120 h was recorded. Compared to CLP-treated animalswith no additional intervention, CR at 5 h in IL-10/ mice and at 5,10, and 15 h in IL-10/ mice led to improved survival. However,intervention in IL-10/ mice at 10 h and in IL-10/ mice at 20 h wasnot successful (FET, P 0.15). ND, not done.

FIG. 6. Treatment with rhIL-10 delayed the onset of lethality andimproved long-term survival in both IL-10/ and IL-10/ mice.Groups of IL-10/ and IL-10/ mice underwent CLP with an 18-gauge needle. At 5 h after CLP-treated mice (IL-10/, dashed line;IL-10/, solid broken line) received 1 g of rhIL-10 in 0.25 ml ofnormal saline (n 20) by subcutaneous injection. Control mice (IL-10/, dotted line; IL-10/, solid thin line) were injected with normalsaline alone (n 30). The time of death within 10-h intervals wasrecorded, and the data were expressed as the cumulative percentage ofmice alive within each interval. Survival of treated mice in both strains,as determined by the Logrank test, was significantly improved com-pared to untreated animals (P 0.007).

TABLE 1. IL-10 Treatment delays the onset of lethality

Mouse straina rhIL-10treatmentb% Survival at 30 h

after CLP Pc

IL-10/ 26 ] 0.002IL-10/ 75 ] 0.94IL-10/ 80 ]IL-10/ 100 0.07

a Each group of mice on a C57BL/6J background underwent CLP with an18-gauge needle.

b Five hours later the mice were injected subcutaneously with 1 g of rhIL-10in 0.25 ml of normal saline (n 20) or vehicle alone (n 30).

c Statistical differences between groups was determined by using Fishers exacttest.

4444 LATIFI ET AL. INFECT. IMMUN.

mice with rhIL-10 at the time of CLP did not lead to a signif-icant improvement in survival (data not shown).

Since the onset of mortality was delayed in IL-10/ andIL-10/ mice treated with rhIL-10 5 h after CLP, we hypoth-esized that the efficacy of rescue surgery would be significantlyimproved (Fig. 7). Treatment of IL-10/ mice with rhIL-10increased the effectiveness of CR at 10 h from a survival rate of20 to 80% (FET, P 0.02). Similarly rhIL-10 treatment inIL-10/ mice significantly improved survival after rescue sur-gery at 20 h from 20 to 90% (FET, P 0.006). Thus, IL-10treatment extends the therapeutic window for rescue surgeryand delays the onset of irreversible shock.

DISCUSSION

In this report we demonstrate that in sepsis triggered by CLPthe transition from reversible sepsis to irreversible septicshock, defined as the time point at which rescue surgery fails,is regulated by IL-10. The absence of IL-10 leads to a morerapid onset of mortality after CLP and an earlier transitionfrom reversible to irreversible sepsis. In contrast, treatment ofmice with rhIL-10 after CLP delays the onset of lethality,improves survival, and extends the therapeutic window forrescue surgery. We therefore propose that IL-10 regulates apoint of no return for recovery from severe sepsis.

The more-rapid onset of mortality in IL-10/ mice com-pared to IL-10/ mice after CLP is similar to data obtained byusing blocking antibodies to IL-10 (28, 31). However, the even-tual survival in anti-IL-10-treated mice was not that muchlower compared to that of control mice and, in the study bySong et al., was not significantly different (28). Similarly, in ourstudy the ultimate lower survival in IL-10/ mice was notsignificantly different to that of IL-10/ mice after eitherlethal or sublethal CLP. Our results, together with these pre-vious reports, indicate that endogenous IL-10 alters the kinet-ics of the septic response in the CLP model but probably doesnot dramatically affect long-term survival.

Increased synthesis of proinflammatory cytokines in IL-

10/ compared to IL-10/ mice after CLP is not unexpectedsince IL-10 is known to suppress the production of TNF- andIL-6 (16). The 15-fold-greater levels of TNF- and IL-6 afterCLP in the IL-10/ mice may contribute to their more rapidonset of mortality. It is unlikely that the more rapid lethality ofIL-10/ mice is due to increased bacteremia, since in anEscherichia coli peritonitis model increased lethality of IL-10/ mice was associated with an accelerated bacterial clear-ance but greater organ dysfunction (27). Therefore, it could bethat the exaggerated proinflammatory cytokine response in theIL-10/ mice after CLP leads to a more rapid onset of irre-versible damage to vital organs and hastens the onset of irre-versible shock.

In the CLP model it is well described that survival can besignificantly improved by the use of antibiotics (18) and fluidresuscitation (18, 32), both treatments being utilized in ourexperimental CLP protocol. Removal of the necrotic cecumafter CLP has also been shown to improve outcome (1). Theresults from our therapeutic approach that involves removal ofthe necrotic cecum at various times after CLP has a strikingclinical correlate. In human disease the mainstay of clinicalmanagement for bacterial peritonitis is to operate and removethe infectious nidus. With timely intervention this leads toresolution of the peritonitis and recovery of the patient in themajority of cases. However, some patients go on to die fromunremitting sepsis and multiorgan failure despite appropriatesurgery and supportive care (26). At the time of presentationthese critically ill patients appear alike in their level of acuity asthose who recover but are presumably further advanced in thedisease process. Similarly, as presented here, the efficacy of CRafter CLP decreases the longer it is delayed, and the transitioninto irreversible shock in both IL-10/ and IL-10/ micedoes not correlate with any obvious clinical signs nor anychanges in the serum cytokine profile. However, our findingthat the therapeutic window for rescue surgery is extended by10 h in IL-10/ compared to IL-10/ mice suggests thatIL-10 may have a significant role in regulating the onset of thepoint of no return and irreversible shock.

If IL-10 does control the onset of irreversible shock in severesepsis, then the treatment of IL-10/ and IL-10/ mice withIL-10 should delay the onset of mortality and lengthen thetherapeutic window for rescue surgery. Our data show that forboth strains of mice treated with rhIL-10 after CLP the onsetof mortality was delayed, and the efficacy of CR was nowextended beyond the time at which it was previously ineffectivein untreated mice. Thus, the combination therapy of rhIL-10and late CR after lethal CLP can lead to increased recovery ofseptic mice.

The modest improvement in survival in IL-10/ mice afterrhIL-10 treatment alone 5 h after CLP is consistent with aprevious report in which a 50% improvement was seen whenIL-10 was given as a single dose 6 h after CLP (17). Similarly,our findings that treatment with rhIL-10 at the time of CLPdoes not increase survival (data not shown) are in agreementwith reports showing that initiation of IL-10 treatment at thetime of CLP does not improve outcome (17, 24). It may be thatIL-10 is not critical at the initiation of the septic response, asevidenced by the ability of early surgical intervention with CRat 5 h to partially reverse CLP-induced lethality in IL-10/

mice. However, soon after the septic response is initiated, the

FIG. 7. rhIL-10 treatment extends the therapeutic window for res-cue surgery after CLP. Groups of IL-10/ and IL-10/ mice receivedeither no treatment (, n 30), CR alone at 10 h or 20 h, respectively(, n 10); or 1 g of rhIL-10 5 h after lethal CLP and CR at 10 hor 20 h (u, n 10). In rhIL-10-treated mice, the efficacy of rescuesurgery by CR was now significantly improved at the time point whenit had previously failed in both strains of mice (FET, P 0.02 []).

VOL. 70, 2002 IL-10 CONTROLS THE ONSET OF IRREVERSIBLE SEPSIS 4445

absence of IL-10 supports a more rapid onset of mortality, yetthe ultimate impact of the absence of IL-10 on long-termsurvival is not that great. In contrast, exogenous IL-10 admin-istered during the reversible phase of sepsis improves outcome.We also hypothesize that administering IL-10 too late duringthe irreversible phase of shock has a limited therapeutic ben-efit, since blocking endogenous IL-10 late after CLP improvessurvival (28). Therefore, therapy with IL-10 once irreversibleshock has begun may be harmful.

Not surprisingly, the role of IL-10 in sepsis is complex, withpotentially opposite effects depending on the timing of inter-vention and whether endogenous versus exogenous IL-10 ismanipulated. The findings presented here, however, help toclarify the confusion by demonstrating that IL-10 functions toregulate the onset of irreversible septic shock and the thera-peutic window for rescue surgery.

ACKNOWLEDGMENTS

We thank Scott Fulton for providing the tribromoethanol anesthetic,David Spencer, Doug Kou, Robin Jump, and Kelly Miller for theirtechnical assistance. Critical discussions and review of the manuscriptby Giovanni Latella, Fred Heinzel, Leslie T. Webster, Jr., and JeffreyL. Blumer are much appreciated.

This study was supported by grants HD-31323 (Jeffrey L. Blumer,Department of Pediatrics, Case Western Reserve University, Cleve-land, Ohio) and CA-77717 (A.D.L.) from the NIH.

REFERENCES

1. Baker, C. C., I. H. Chaudry, H. O. Gaines, and A. E. Baue. 1983. Evaluationsof factors affecting mortality rate after sepsis in a murine cecal ligation andpuncture model. Surgery 94:331335.

2. Balk, R. A. 2000. Severe sepsis and septic shock: definitions, epidemiology,and clinical manifestations. Crit. Care Clin. 16:179192.

3. Berg, D. J., K. Kuhn, K. Rajewsky, W. Muller, S. Menon, N. Davidson, G.Grunig, and D. Rennick. 1995. Interleukin-10 is a central regulator of theresponse to LPS in murine models of endotoxic shock and the Shwartzmanreaction but not endotoxin tolerance. J. Clin. Investig. 96:23392347.

4. Beutler, B., I. W. Milsark, and A. C. Cerami. 1985. Passive immunizationagainst cachectin/tumor necrosis factor protects mice from lethal effect ofendotoxin. Science 229:869871.

5. Cain, B. S., D. R. Meldrum, A. H. Harken, and R. C. McIntyre. 1998. Thephysiologic basis for anticytokine clinical trials in the treatment of sepsis.J. Am. Coll. Surg. 186:337351.

6. Damas, P., J.-L. Canivet, D. de Groote, Y. Vrindts, A. Albert, P. Franchi-mont, and M. Lamy. 1997. Sepsis and serum cytokine concentrations. Crit.Care Med. 25:405412.

7. Despond, O., F. Proulx, J. A. Carcillo, and J. Lacroix. 2001. Pediatric sepsisand multiple organ dysfunction syndrome. Curr. Opin. Pediatr. 13:247253.

8. Doughty, L., J. A. Carcillo, S. Kaplan, and J. Janosky. 1998. The compen-satory anti-inflammatory cytokine interleukin 10 response in pediatric sepsis-induced multiple organ failure. Chest 113:16251631.

9. Doughty, L. A., S. S. Kaplan, and J. A. Carcillo. 1996. Inflammatory cytokineand nitric oxide responses in pediatric sepsis and organ failure. Crit. CareMed. 24:11371143.

10. Eskandari, M. K., G. Bolgos, C. Miller, D. Nguyen, L. E. DeForge, and D. G.Remick. 1992. Anti-tumor necrosis factor antibody therapy fails to preventlethality after cecal ligation and puncture or endotoxemia. J. Immunol.148:27242730.

11. Friedman, G., S. Jankowski, A. Marchant, M. Goldman, R. J. Kahn, andJ. L. Vincent. 1997. Blood interleukin 10 levels parallel the severity of septicshock. J. Crit. Care 12:183187.

12. Gazzinelli, R. T., M. Wysocka, S. Hieny, T. Scharton-Kersten, A. Cheever, R.

Kuhn, W. Muller, G. Trinchieri, and A. Sher. 1996. In the absence ofendogenous IL-10, mice acutely infected with Toxoplasma gondii succumb toa lethal immune response dependent on CD4 T cells and accompanied byoverproduction of IL-12, IFN- and TNF-. J. Immunol. 157:798805.

13. Group, T. A. S. S. 2001. Randomized, placebo-controlled trial of the anti-tumor necrosis factor antibody fragment afelimomab in hyperinflammatoryresponse during severe sepsis: the RAMSES study. Crit. Care Med. 29:765769.

14. Hollenberg, S. M., A. Dumasius, C. Easington, S. A. Colilla, A. Neumann,and J. E. Parrillo. 2001. Characterization of a hyperdynamic murine modelof resuscitated sepsis using echocardiography. Am. J. Respir. Crit. CareMed. 164:891895.

15. Howard, M., T. Muchamuel, S. Andrade, and S. Menon. 1993. Interleukin 10protects mice from lethal endotoxemia. J. Exp. Med. 177:12051208.

16. Huhn, R. D., E. Radwanski, S. M. OConnell, M. G. Sturgill, L. Clarke, R. P.Cody, M. B. Affrime, and D. L. Cutler. 1996. Pharmacokinetics and immu-nomodulatory properties of intravenously administered recombinant humaninterleukin-10 in healthy volunteers. Blood 87:699705.

17. Kato, T., A. Murata, H. Ishida, H. Toda, N. Tanaka, H. Hayashida, M.Monden, and N. Matsuura. 1995. Interleukin 10 reduces mortality fromsevere peritonitis in mice. Antimicrob. Agents Chemother. 39:13361340.

18. Lundblad, R., and K.-E. Giercksky. 1995. Effect of volume support, antibi-otic therapy, and monoclonal antiendotoxin antibodies on mortality rate andblood concentrations of endothelin and other mediators in fulminant intra-abdominal sepsis in rats. Crit. Care Med. 23:13821390.

19. McMasters, K. M., J. C. Peyton, D. J. Hadjiminas, and W. G. Cheadle. 1994.Endotoxin and tumor necrosis factor do not cause mortality from caecalligation and puncture. Cytokine 6:530536.

20. Meduri, G. U., S. Headley, G. Kohler, F. Stentz, E. Tolley, R. Umberger, andK. Leeper. 1995. Persistent elevation of inflammatory cytokines predicts apoor outcome in ARDS, plasma IL-1 and IL-6 levels are consistent andefficient predictors of outcome over time. Chest 107:10621073.

21. Murphy, K., S. B. Haudek, M. Thompson, and B. P. Giroir. 1998. Molecularbiology of septic shock. New Horiz. 6:181193.

22. Ohlsson, K., P. Bjork, M. Bergenfeldt, R. Hageman, and R. C. Thompson.1990. Interleukin-1 receptor antagonist reduces mortality from endotoxinshock. Nature 348:550552.

23. Papaioannou, V. E., and J. G. Fox. 1993. Efficacy of tribromoethanol anes-thesia in mice. Lab. Anim. Sci. 43:189192.

24. Remick, D. G., S. J. Garg, D. E. Newcomb, G. Wollenberg, T. K. Huie, andG. L. Bolgos. 1998. Exogenous interleukin-10 fails to decrease the mortalityor morbidity of sepsis. Crit. Care Med. 26:895904.

25. Remick, D. G., D. E. Newcomb, G. L. Bolgos, and D. R. Call. 2000. Com-parison of the mortality and inflammatory response of two models of sepsis:lipopolysaccharide versus cecal ligation and puncture. Shock 13:110116.

26. Riche, F. C., B. P. Cholley, Y. H. Panis, M.-J. C. Laisne, C. G. Briard, A.-M.Graulet, J. L. Gueris, and P. D. Valleur. 2000. Inflammatory cytokine re-sponse in patients with septic shock secondary to generalized peritonitis.Crit. Care Med. 28:433437.

27. Sewnath, M. E., D. P. Olszyna, R. Birjmohun, F. J. ten Kate, D. J. Gouma,and T. van der Poll. 2001. IL-10 deficient mice demonstrate multiple organfailure and increased mortality during Escherichia coli peritonitis despite anaccelerated bacterial clearance. J. Immunol. 166:63236331.

28. Song, G. Y., C.-S. Chung, I. H. Chaudry, and A. Ayala. 1999. What is the roleof interleukin 10 in polymicrobial sepsis: anti-inflammatory agent or immu-nosuppressant? Surgery 126:378383.

29. Sullivan, J. S., L. Kilpatrick, J. Costarino, A. T., S. C. Lee, and M. C. Harris.1992. Correlation of plasma cytokine elevations with mortality rate in chil-dren with sepsis. J. Pediatr. 120:510515.

30. van der Poll, T., R. Malefyt, S. M. Coyle, and S. F. Lowry. 1997. Antiinflam-matory cytokine responses during clinical sepsis and experimental endotox-emia: sequential measurements of plasma soluble interleukin (IL)-1 receptortype II, IL-10, and IL-13. J. Infect. Dis. 175:118122.

31. van der Poll, T., A. Marchant, W. A. Buurman, L. Berman, C. V. Keogh,D. D. Lazarus, L. Nguyen, M. Goldman, L. L. Moldawer, and S. F. Lowry.1995. Endogenous IL-10 protects mice from death during septic peritonitis.J. Immunol. 155:53975401.

32. Wilson, M. A., M. C. Chou, D. A. Spain, P. J. Downard, Q. Qian, W. G.Cheadle, and N. Garrison. 1996. Fluid resuscitation attenuates early cyto-kine mRNA expression after peritonitis. J. Trauma 41:622627.

Editor: A. D. OBrien

4446 LATIFI ET AL. INFECT. IMMUN.