t¿

Cellular mechanisms of abscess formation:Macrophage procoagulant activity and

majar histocompatibility complexrecognitionRobert G. Sawyer, MD, and Timothy L. Pruett, MD, Ann Arbar, Mich., andCharlottesville, Va.

Background. Macrophage procoagulant activity (PCA) has been proposed as a key mediator of abscessformation. Experimentally, transient systemic bacterial infections lead to increased numbers ofintraabdominal abscesses after a subsequent episode of peritonitis. We tested the hypothesis that these

events were regulated by classic majar histocompatibility complex (MHC)-restricted antigen processingand presentation to lymPhocytes followed by lymphocyte-mediated up-regulation of macrophage PCA.Methods. In viera, macrophages and lymPhocytes from BALB/c or C57BL/6 mice either untreated orpreexposed to Escherichia coli were coincubated with bacteria or lipopolysaccharide. Celllysates weretested for PCA in a one-step clotting assay. In vivo, mice were either preexposed to E. coli or receivedpassive transfer of lymPhocytes from MHG-compatible or MHG-incompatible and naive or preexposeddonors; peritonitis and intraabdominal abscesses were afterwards induced with E. coli, Bacteroidesfragilis, and a sterile fecal adjuvant. Mice were killed after 10 days and were studied for abscess

number and bacterial composition.Results. The presence of lymPhocytes consistently increased macrophage PCA,. lymPhocytes frompreexposed donors induced twice as much PCA as lymPhocytes from naive donors regardless of MHCbackground. Both bacterial preexposure and passive transfer of lymPhocytes from preexposed donorsincreased later intraabdominal abscess number in an MHG-restricted fashion.Conclusions. Transient infections enhance subsequent lymphocyte-mediated macrophage PCA,correlating with in creases in abscess formation after peritonitis. The need for MHC identity toreproduce these results via passive transfer in vivo is consistent with classic T -cell receptor-mediatedantigen presentation and lymPhocyte aétivation before enhancement of PCA during peritonitis. (Surgery

1996,.120:488-95.)

From the Deparlment of Surgery, The University of Michigan, Ann Arbor, Mich., and theDeparlments of Surgery and lnternal Medicine, The University of Virginia, Charlottesville, Va.

RElA11VELY TRANSIENT INFECTlONS such as catheter-re-lated or wound sepsis are cornmon among criticalIy ilI

patients undergoing operation. The immunologic ef-

fects on subsequent infections after such a brief expo-

sure to microbes and their associated antigens remainunclear. To study this question we have developed a

murine model ofbacteremia followed by peritonitis and

intraabdominal abscess formation. Preexposure of miceto 103 live Escherichia coli 2 and 1 weeks before the

induction of peritonitis with E. coli, Bacteroides fragilis,

and autoclaved stool adjuvant leads to a significant in-crease in fue number of abscesses formed 1 whether fue

injection is given by intraperitoneal, intracardiac, orsubcutaneous routes.2 This response is reproduciblewith preexposure to other gram-negative nosocomialpathogens but not lipopolysaccharide (LPS) alone,gram-positive bacteria, or fungi.3 Passive transfer exper-iments have shown that this effect is dependent on

CD4+ T lymphocytes.4One possible explanation for this response is in-

creased fibrin deposition mediated by macrophage-de-rived procoagulant activity (PCA). PCA-induced fibrindeposition is critical to many forms of inflammation in-cluding peritonitis.5.6 The importance of fibrin inabscess formation is supported both histologically by thepresence of a fibrin rim surrounding a core of bacteriaand white cells 7 and functionally b~. fue observation thatinterference with fue generation of fibrin by heparin,S. 9

Supported by University ofVirginia Research and Development grant

no. 5-507-RR05431-28.

Accepted for publication Feb. 8, 1996.

Reprint requests: Robert G. Sawyer, MD, 1500 E. Medical Center Dr.,Department of Surgery, Division of Transplantation, 2926 Taubman

0331, Ann Arbor, MI 48109.0331.

Copyright @ 1996 by Mosby-Year Book, lnc.0039-6060/96/$5.00 + O 11/56/72754

488 SURGERY

Sawyer and Pruett 489Su rgeryVolume 120, Number J

aprotinin;o or tissue plasminogen activatorll lessensfue severity of peritonitis and prevents experimentalabscesses. A central role for PCA in the immuneresponse leadingto abscesses has thus been proposed.12It is interesting that in our model fue ability of nosoco-mial pathogens to up-regulate fue number of subse-quent abscesses parallels their induction of macrophagePCA in vitro,3 and in vivo reduction of peritonealmacrophage PCA with lipids correlates with decreasedintraabdominal abscess formation.13 In addition,T lymphocytesl4-17 and more specifically CD4+ cellslS;21and ThO and Thl clones22 can significantly enhancemacrophage PCA, probably via monocyte/macro-phage procoagulant inducing factor (MPIF)23-25 orother cytokines such as interferon-'Y26,27 or interleu-kin-2.28 Although fue necessity for classic T <en recep-tor-mediated, major histocompatibility complex(MHC)-restricted antigen presentation to fuese lym-phocytes is less clear, many systems involving LPS or an-tigen-stimulated PCA require MHC recognition.29-WTaken as a whole, fuese data suggest that fue increasein abscess number in our system is mediated by a CD4+lymphocyte-driven, MHG-restricted up-regulation ofmacrophage PCA. To test this hypothesis we studiedbacterial preexposure-induced, lymphocyte-mediatedchanges in both ex vivo macrophage PCA and in vivoabscess formation under MHG-identical and MHC-di~similar conditions.

MATERIAL AND METHODS

Animals. Male Balb/c or C57BL/6 mice weighing 20to 25 gm (Charles River Laboratories, Wilmington,Mass.) were used for all experiments. Animals werehoused in stainless steel wire cages and received PurinaLab Chow (Purina MilIs, Sto Louis, Mo.) and water ad li-bitum according to National Research Council stan-dards. AlI experiments were approved by fue Universityof Virginia Animal Use Committee.

Bacteria, LPS, and fecal adjuvant. E. coli 0065 (018,K-, H-nonmotile), a clinical isolate, was grown from asingle colony off a stock plate by incubating in brain-heart infusion broth (Becton Dickinson and Co., Cock-eysville, Md.) for 18 hours in a 370 C oscillating waterbath. Cultures were centrifuged for 12 minutes at 2400gat 40 C, and pellets were washed twice withO.9% salinesolution. Mter fue second wash, bacteria were quanti-fied with serial dilutions and plate enumeration. B. fra-gilis (ATCC 23745) was prepared by inoculating 1 mlofa stock chopped meat broth solution into brain-heartinfusion broth supplemented with hemin and menadi-one (Carr-Scarborough, Stone Mountain, Ga.) and in-cubated anaerobically for 18 hours at 370 C. Cultureswere centrifuged for 12 minutes at 2400g at 40 C, andpellets were washed twice in ,prereduced Hanks' bal-anced salt' solution (Sigma Chemical Company, Sto

Louis, Mo.). Mter washing was done, bacteria were se-rially diluted and quantified by plate dilution on bru-cella agar (Becton Dickinson) in a Coy anaerobe cham-ber (Coy Manufacturing, Ann Arbor, Mich.). LPS 0:26B6 (Sigma Chemical Company) was used as a standardPCA stimulant. ..,

Sterile fecal adjuvant solution was made by grindingdried mouse reces joto afine powder, adding to 0.9%saline solution to form a 5% (wt/vol) mixture, and ster-ilizingin asteam autoclave at 1210 C for 20 minutes. Thispreparation was refrigerated at 40 C, and sterility wasensured before each use by negative culture on blood

agar plates.Preexposure procedure and induction and harvest-

ing of peritoneal macrophages and splenic lympho-cytes. When indicated, mice were inoculated with 103live E. coli vía intraperitoneal injections (total volume,100 pl) 2 and 1 weeks before harvesting of peritonealmacrophages or splenic lymphocytes was performed.Neither the control group nor animals to be killed as asource of naive lymphocytes received pretreatment.Previous experiments had shown that intraperitonealinjections of vehicle (saline solution or prereducedHanks' balanced salt solution) in a manner analogousto inoculations with live bacteria did not affect later in-traabdominal abscess formation. No deaths resultedfrom fuese inoculations, and examination of animalstreated in this fashion has revealed no intraabdominalpathologic condition and sterile cultures of the perito-neum, liver, and spleen.4

To obtain peritoneal macrophages, mice underwentintraperitoneal injection with 2 mI Brewer's thioglycol-late (Difco Co., Detroit, Mich.) 2 days before beingkilled for harvest of cells. At death mice had their abdo-mens prepared sterilely and received intraperitonealinjections of 2 mI ice-cold Roswell Park Memorial Insti-tute medium (RPMI) 1640 (Sigma Chemical Compa-ny). The lavage fluid was removed aseptically, andresultant cells were washed three times in media at 40 Cat 500gfor 10 minutes and were resuspended in RPMI-1640 with 10% fetal bovine serum (Sigma ChemicalCompany). Cells treated this way were found to be con-sistently greater than 93% macrophages by nonspecificesterase staining and had greater than 85% viability bytrypan blue exclusion.

Mter peritoneal macrophages were removed, spleenswere removed and placed in RPMI-l640 at 40 C, andcells were gently teased out and ron across Lym-pholyte-M separation media (Cedarlane Laboratories,Homby, Ontario, Canada) at 250 C at 500gfor 20 min-utes to remove red cells. Pellets were resuspended inRPMI-l640, and cells were allowed to adhere to polysty-rene tubes for 2 hours at room temperature to removemacrophages. Nonadherent cells were washed threetimes in medium at 40 C at 500gfor 10 minutes and re-

SurgerySeptember 1996

490

Sauryer

and Pruett

by halothane overdose. Their abdomens were sterilelyopened, and all abscesses 2 mm or greater were excised,measured in greatest diameter, placed in prereducedHanks' balanced salt solution, and weighed. Prevíousexperiments with this model have shown that abscessesless than 2 mm in diameter are invariably sterile(unpublished data). Abscesses were minced, and fuehomogenate was serially diluted and quantitativelyplated aerobically on McConkey's agar and anaerobi-cally on brucella agar. Surveillance culture for non-gram-negative organisms was carried out intermittendyvía aerobically incubated blood agar plates. Growth wasevaluated 24 and 48 hours after plating. Minimum de-tectability for bacteria was =5 x 102 colony-forming

units/abscess.Twelve groups (six BALB/c and six C57BL/6) were

studied for mortality and abscess formation. Infectedanimals were either not pretreated (naive) or preex-posed to 103 live E. coli or received fue passive transferof 1 07lymphocytes from naive or E. coli preexposed cellsBALB/c or C57BL/6 mice.

Statistics. Mortalities were compared by chi-squaredanalysis, and multiple means were compared by analy-gis of variance (Fisher protected least significant difIer-ence); a p value less than 0.05 was considered signifi-canto Statistical software packages (StatWorks, Cricket-Software, Philadelphia, Pa., and StatView512+, AbacusConcepts, Calabasas, Calif.) were used for calcula-tions.

suspended in RPMI-1640 with 10% fetal bovine serum.Cells treated this way were found to be consistently lessthan 4% macrophages by nonspecific esterase stainingand had greater than 85% viability by trypan blueexclusion. Wright's staining revealed greater than 90%lymphocytes. Of note, previous work has shown little orno activity of naive or immunized B lymphocytes in ourabscess model.4

When indicated, 107 splenic lymphocytes in 150 plRPMI-1640 were passively transferred via intracardiacinjection under aseptic conditions to recipient miceunder light halothane anesthesiajust before fue induc-tion of peritonitis. This procedure incurred an acutemortality of less than 5%.

Macrophage procoagulant activity assay. 106 macro-phages, 107lymphocytes, or both were coincubated inpolypropylene tubes under 5% CO2 at 370 C for 8 hourswith nothing, 103live E. coli, or 5 pg LPS. Mter incuba-tion was performed, cells were resuspended, washedthree times in RPMI-1640 at 40 C at 500gfor 10 minutes,lysed by three freeze-thaw cycles at -800 C, and storedat -800 C until assay.

PCA was measured in terms of the reduction in timeneeded to form a clot in aone-step clotting assay as pre-viously described.I2 In brief, 0.1 mI of freeze-thaw lysatesample or rabbit thromboplastin standard (AmericanHospital Corporation, Dade, Fla.) was mixed with 0.1 mIcitrated human plasma and 0.1 mI 25 mmol/L CaCI2,placed in a water bath at 370 C, and observed for clotformation. AlI samples were ron in duplicate. A stan-dard curve of PCA units was constructed with 37.5mg/ml (total, 3.75 mg) of thromboplastin standarddesignated as 100,000 mU PCA; plotting fue logIa ofPCA units versus time in seconds to form a clot resultedin a linear relationship (R2 value typically >0.97). Con-version of time to clotting to relative PCA units for sam-pIes was performed with this relationship and statisticalsoftware.

Twenty groups were studied for PCA (Table 1) withfour sources of macrophages (106 naive or E. coli preex-posed cells from BALB/c or C57BL/6 mice) and fivesources of coincubated lymphocytes (none, or 107 cellsfrom naive or E. coli preexposed BALB/c or C57BL/6

mice).Induction and analysis of abscesses. Abscesses were

induced as previously described.I Mter inhalational an-esthesia with halothane, mice received an intraperito-neal injection of a mixture of 103 live E. coli, 105 live B.fragilis, and 50 pl sterile fecal adjuvant solution (totalvolume, 175 pl). Acute volume resuscitation was givenvia subcutaneous injection of 0.5 mI 0.9% saline solu-tion, and animals were retumed to their cages and al-lowed oral intake offood and water ad libitum. Ten daysafter abscess induction was performed, mice were killed

RESULTSProcoagulant activity. Unstimulated PCA for macro-

phagesfromnaive (non-preexposed) animalswas53 :t 9and 17 :t 2 mU per 106 cells for BALB/c and C57BL/6mice, respectively; LPS-stimulated PCAfor macrophagesfrom naive mice was 380:t 67 and 560 :t 110 mU per106 cells for BALB/c and C57BL/6 strains. LPS-stimu-lated PCA measured for 107 isolated lymphocytes fromnaive animalswas 105:!: 11 and 63:!: 8mU forBALB/cand C57BL/6, respectively; E. coli-stimulated PCA from107 BALB/c and C57BL/6 lymphocytes was 260 :!: 38and63:t 8fornaivedonorsand350:!: 110and150:t 22for preexposed donors, respectively. This small amoun tofPCA measured for isolated lymphoCytes was probablydue to contaminating macrophages.

PCA for various combinations of macrophages andlymphocytes from naive and E. coli preexposed donorsof fue two strains are given in Table l. Of note, fue ad-dition oflymphocytes to macrophages always increasedPCA. Preexposure of lymphocyte donors to E. coli uni-forrnly led to an increased PCA when compared withsimilar groups using naive lymphocyte donors. Preex-posure of donar macrophages had little effect, exceptwhen BALB/ c macrophages were coincubated either

Sauryer and Pruett 491SurgeryVolume 120, Number

Table l. Procoagulant activity of macrophages in response to E. coli by origin of coincubated lymphocytes

Lymphocyte SOUTCe

PCA expressed in milliunits = SEM, per 106 macrophages tested: N = 5 per group; Statistics by anaIysis ofwriance (Fisher P~D)../J S 0.02 versus no Iymphocytes added, same macrophage source.tp s 0.02 versus naive Iymphocytes from same mouse strain, same macrophage source.tps 0.02 versus naive macrophages from same mouse strain, same Iymphocyte source.§p s 0.02 versus macrophages from other mouse strain with same donor treatment, same Iymphocyte source.

with no or C57BL/6 preexposed lymphocytes. Whencontrolled for preexposure status and macrophagesource tested, lymphocytes from both mouse strainswere equally effective in up-regulating PCA, even cros&-ing MHC liDes. BALB/c macrophages had significantlyhigher PCA than C57BL/6 macrophages when coincu-bated with preexposed lymphocytes from either mousestrain. The importance of this lone strain difference isnot clear.

In vivo experlments.Table 11 shows groupsize, mor-tality, and the number ofbacteria recovered per abscessfrom lO-day survivors for the 10 in vivo groups studied.AlI deaths occurred within 72 hours of initiation ofperitonitis, usually between 24 and 48 hours of infec-tion. Surveillance cultures revealed no bacteria otherthan E. coli and B. fragilis. None of fue variablespresented in Table 11 are significantly different betweengroups (p> 0.02). No significant difference was seen inabscess size or weight benveen groups.

The number of abscesses found at death at dar 10 isshown in Fig. l. Simple preexposure of animals to 103E. coli led to significantly greater numbers of abscessesin both strains, reconfirming our earlier work andshowing this phenomenon in C57BL/6 mice. Passivetransfer of 107 lymphocytes from preexposed same-strain donors to naive siblings reproduced this resultoThe passive transfer of lymphocytes from naive donorsor from preexposed, dissimilar strain donors had no ef-fect on abscess number.

DISCUSSION

Macrophage PCA has been proposed to be a criticalelement in the host response to both systemic sepsis31. 32and localized infection.12. 33, 34 The studies reported

here attempt to address via both ex vivo measurementof PCA and in vivo experiments the cellular control ofup-regulated PCA in the enhanced intraabdominal ab-scess formation seen after transient bacterial infec-tions.l '-

Several immunoregulatory mechanisms have beendescribed for PCA induction in vitro, depending pri-marily on the nature of fue PCA stimulus. First, alloan-tigen/ mixed lymphocyte reaction-induced PCAappearsto involve two distinct pathways: one mediated by lym-

phocyte-derived monocyte/macrophage procoagulantinducing factor (MPIF) and other cytokines, and fuesecond regulated by direct lymphocyte-macrophagecontact.19. 21, 35. 36 The key lymphocytes arrear to be

CD4+ and have either a ThO or Thl cytokine secretionpattem.21, 22 Neither pathway is MHG-restricted.21

Second, induction of PCA via LPS,14,17,19 immunecomplexes,15,37 or concanavalin A24,25 appears to in-volve a more direct stimulation of monocytes/macro-phages and does not always require fue presence oflymphocytes.12. 38, 39 Lymphocyte help, when detected,

mar be via MPIF23 and other cytokines such as interfer-on-'Y,26, 27 is MHG-restricted, and is generally not de-

pendent on antigen processing and presentation tolymphocytes.15.16.29 Other more complicated modelsinvolve PCA induced by complex antigens such as oval-bumin,18 fue purified protein afilie tubercle bacillus,30bacteria,3, 12, 40 or viruses.3, 3S, 40 Lymphocyte help is

variable in fuese systems and depends on the exact na-ture of the organism or protein stimulating PCA, (forexample, live versus dead organisms, bacteria versus vi-ruses, specific bacterial species, and inoculum size) .s, 40

When studied in these models, lymphocyte help isMHG-restricted and dependent on antigen presenta-tion by monocytes,20. so findings in accord with fue clas-

sic immune response to antigen. It is suspected but un-proven that cytokines including MPIF are still involvedin the lymphocyte-macrophage communication under,these conditions.

The ex vivo PCA experiments described here, whentaken alone, support fue hypothesis that this function iscritical to fue changes in abscess formation seen aftertransient bacterial infection in our modelo Even thoughcoincubated lymphocytes from all sources enhanced

Sawyer and Pruett 493SurgeryVolume 120, Number

vivo assay, however, does not support this conceptoThere are several possible explanations for this out-come. The PCA mar have been induced thro\lgh an al-loantigen response similar to that previously describedand lacking MHC restriction,21, 35, 36 although this ex-

planation is unlikely, because alloreactive PCA typicallydoes not peak until at least 48 hours after fue initiationof a mixed lymphocyte reaction (MLR),35-37 and ourcoincubation period was only 8 hours. Inaddition, evenif therewas some aIloantigen-induced PCA, it could notaccount for the differences seen between lymphocytesfrom naive and preexposed donors, because they would

presumably represent equivalent alloantigenic targets.It is possible that in the small incubation volume used,macrophages were able to activate lymphocytes indi-rectly without antigen presentation (for example víaadequate accumulation of tumor necrosis factor-a, in-terleukin-l,41 or some other monokine). Altematively,LPS from macrophage-mediated bacteriolysis marhave been sufficient to stimulate lymphocytes to pro-duce MPIF and other cytokines and further up-regulatePCA.

The inconsistent finding of increased PCA from pre-exposed rather than naive macrophage donors is ofunknown importance. The greater PCA that under-went assay from BALB/c-derived macrophages com-pared with C57BL/6 donors in some groups underidentical conditions is undoubtedly due to mouse straindifferences. It is interesting to note that in onemodel of murine hepatitis type 3, macrophages fromBALB/ c mice had a particularly strong PCA response tovirus,33 whereas in another model of alloantigen-stimu-lated PCA, C57BL/6-derived macrophages exhibitedPCA significantIy greater than those from BALB/cmice.21

The lymphocyte passive transfer data show fue centralrole of lymphocytes in intraabdominal abscess forma-tion and its up-regulation after transient bacterial infec-tion in C57BL/6 mice. Even though isolated CD4+ Tcells were not used in these experiments, there is littIereason to doubt they are the criticallymphocyte subsetas previously shown for BALB/c mice.4 The similarmortality and abscess bacteriologic characteristicsamong groups are also consistent with our previouswork, implying again that fue immunologic mecha-nisms controIling these variables differ from those thatdetermine abscess number. The exact nature of theseinteractions is unknown. Although fue importance ofPCA up-regulation, perhaps by MPIF, is emphasized inthis study, the involvement of proinflammatory cyto-kines such as tumor necrosis factor and interleukin-land perhaps even alterations in macrophage function(e. g., increased phagocytosis) are also likely. Studies todelineate these details such as eytokine/CD4 phenotyp-ing (Thl versus Th2) of fue involvedlymphocytes are

currently undeIWay. In addition, an altemative hypoth-esis to account for these findings relates to a possiblecryptic infection after inoculation despite sterile cul-tures from preexposed animals. Intraphagocytic seques-tration has not been completely ruledout; fue passivetransfer of fuese cells could explain fue immunomodu-lating effects noted.

Although consistent with certain models of PCA reg-ulation,30 fue MHC restriction noted in fuese passivetransfer experiments was not predicted by our in vitroPCA results. The reasons for this disparity are unclear.It is possible that, similar to fue PCA data, MHGdissim-ilar lymphocytes could in crease abscess number but thatpassively transferred allogeneic cells are destroyed orrejected before theyare able to up-regulate PCA in vivo.This hypothesis is questionable given the amount oftime (days) needed to fully mount an allogeneic re-sponse (assuming no unexpected presensitization be-tween strains oflaboratory-bred mice) and fue fact thatabscesses in this model are found 4 days after infection(unpublished data) and mar even be established beforethat. It is possible that PCA is not important in the res-olution of peritonitis leading to abscesses, but thishypothesis still seems improbable given the well-de-scribed role of fibrin in peritonitis and abscess forma-tion,41-43 fue clinical44 and experimental association ofB. fragilis with abscesses45.46 and PCA induction,12.40and previous findings that the inhibition of peritoneal

macrophage PCAcorrelates with decreased abscess for-mation.13 The discrepancy between PCA and abscessdata líes most likely in the oversimplification of fue invitro model compared with in vivo responses. Forexample, the roles ofboth polymorphonuclear neutro-phils and noncellular debris, both ubiquitous in ab-scesses! are unaccounted for in this two-cell system.Proof of a central position of PCA to both abscess for-mation and fue changes seen after transient bacterialinfections therefore will depend on fue discovery and invivo use of either a monoclonal antibody to tissue fac-tor that diminishes PCA without depleting monocytes orotherwise altering the immune response or some otherhighly specific inhibitor of PCA.

With a murine system of increased intraabdominalabscess formation after transientbacterial infection, wehave attempted to analyze the importance of MHC re-striction and macrophage PCA in fue cellular regulationof abscess development. Although the ability of lym-phocytes from preexposed donors to up-regulate PCA.ex vivo corresponded to increased numbers of abscessesin vivo, the disparity in the need for MHC identity indi-cates subtle differences between fue two systems. Givenfue pre\'ailing paradigm of the immunologic responseto antigen, however, a classic, MHC-restricted, CD4+ T-cell-mediated enhancement ofmacrophage PCA is stillfue hypothesis most consistent with fue data presented.

SurgerySepternber 1996

494 Sawyer and Pruet

Definitive proof of this supposition awaits both better invitro models and finer tools to dissect in vivo fue com-plex local and systemic interactions that occur during

peritonitis.

REFERENCES

l. Sawyer RG, Adams RB. Spengler MD. Pruett TL. Pre-exposure ofthe peritoneum to live bacteria increases later mixed intraab-dominal abscess formation and alters mortality.] Infect Dis

1991;163:664-7.2. Sawyer RG, Adams RB. Spengler MD. Pruett TL. Transient and

distant infections alter later intraperitoneal abscess formation.

Arch Surg 1991;126:164-8.3. Sawyer RG, Pruett TL. Nosocomial bacteremia-induced increases

in abscess formation correlates with in \Ítro upregulation ofmacrophage procoagulantacti\Íty. CritCare Med 1995;23: 1554-9.

4. Sawyer RG, Adams RB. May AK, Rosenlof LK, Pruett TL. CD4+T cells mediate pre-exposure induced increases in murine intra-abdominal abscess formation. Clin Immunol Immunopathol

1995;77:82~.5. Almdahl SM, Osterud B. Experimental gram-negative peritonitis:

decreased thromboplastin acti\ity in organs with a simultaneousrige of thromboplastin in blood monocytes and peritoneal mac-

rophages. Res EXp Med 1986;186:317-24.6. Almdahl SM. Brox ]H. Osterud B. Mononuclear phagocyte

thromboplastin and endotoxin in patients with secondary bacte-

rial peritonitis. &and] GastroenteroI1987;22:914-8.7. Madri ]A. Inflammation and healing. In: Kissane M. editor.

Anderson's pathology. 9th ed. Philadelphia: WB Saunders Co.

1990:67-110.8. Hau T. Simmons RL. Heparin in the treatment of experimental

peritonitis. Ann Surg 1978;187:294-8.9. Chalkiadakis G. Kostakis A. Karayannacos PE. et al. The effect of

heparin upon fibrinopurulent peritonitis in rats. Surg Gynecol

Obstet 1983;157:257-60.10. Chalkiadakis G. Kostakis A. Karydakis P. et al. Effect of aproti-

nin on fibrinopurulent peritonitis in rats. Arn] Surg 1985;150:

550-3.11. Rotstein OD, Kao]. Fibrinolysis using recombinant tissue plas-

minogen acti\'ator prevents intraabdominal abscesses.] Infect Dis

1988;158:766-72.12. Rosenthal GA, Lery G, Rotstein OO. Induction of macrophage

procoagulant activity by Bacteroides fragilis. 1nfect Immun 1989;

57:338-43.13. Sawyer RG, Pruett TL. 1nhibition of macrophage procoagulant

activity with lipids reduces intraabdominal abscess formation in

mice. Eur Surg Res 1995;27:222-6.14. Edwards RL, Rickles FR The Tole of human T cells (and T cell

products) for monocyte tissue factor generation. ] Immunol

1980;125:606-9.15. Schwartz BS, Edgington TS. Lymphocyte collaboration is re-

quired for induction of murine monocyte procoagulant activityby immune complexes.] ImrnunoI1981;127:438-43.

16. Lery GA, Edgington TS. Lymphocyte cooperation is required foramplification of macrophage procoagulant activity.] Exp Med

1980;151:1232-44.17. Lando fA, Edgington TS. Lymphoid procoagulant response to

bacterial endotoxin in rats. Infect Immun 1985;50:660-6.18. Geczy CL,Farram E, Moon OK, Meyer PA, McKenzie IFC. Mac-

rophage procoagulant activity as a measure of cell-mediated irn-

munity in the mouse.] ImmunoI1983;130:2743-9.19. Gregory SA, Edgington TS. Tissue factor induction in hu-

rnan monocytes: tWO distinct mechanisms displayed by differ-ent alloantigen responsive T -cell clones.] Clin Invest 1985;76:

2440-5.

20. Schwartz BS. Reimauer PJ. HankJA, Sondel PM. Ahuman T cellclone that mediates the monocyte procoagulant response to spe-

cific sensitizing antigen. J Clin Invest 1985;76: 1279-82.

21. Fan SOT. Edgington TS. Clonalanal}"SisofmechanismsofmurineT helper cell collaboration \vith effector cells of macrophage lin-

eage.J ImmunoI1988;141:1819-27.22. Fan So T. Glasebrook AL, Edgington TS. Clonal analysis of CD4+

T helper cell subsets that induce the monocyte procoagulant re-

sponse. Celllmmunoll990;128:52-62.23. Gregory SA, Kombluth RS, Helin H, Remgold HG. Edgington

TS. Monocyte procoagulant inducing factor: a Iyrnphokineinvolved in the T cell-instrllcted monocyte procoagulant re-

sponse to antigen.J ImmunoI1986;137:3231-9.24. R};tn J. Geczy CL. Characterization and purification of mouse

macrophage procoagulant-inducing factor. J Immunol 1986;

137:2864-70.25. R};tnJ. Geczy CL. Macrophage procoagulant inducing factor. In

vivo properties and chemotactic acti,ity for phagocytic cells. J

ImmunoI1988;141:2110-7.26. Moon DK, Geczy CL. Recombinant IFN-'( synergizes with Ii-

popolysaccharide to induce macrophage membrane procoagu-

lants.J IrnmunoI1988;141:153642.27. Schwager l,jungi TW. Effect ofhuman recombinant cytokines on

the induction of macrophage procoagulant activity. Blood

1994;83:152-60.28. Carlsen E. Prydz H. Acti\'ation of monocyte~more than one

process: differential effects of cytokines on monocytes. Scand J

ImmunoI1988;27:401-4.29. Levy GA. Edgington TS. The majar histocompatibility complex

requirement for cellular collaboration in the murine Iyrnphoidprocoagulant response stimulated by bacterial lipopolysaccha-

ride.J ImmunoI1981;128:1284-8.30. Schwartz BS. Antigen-induced monocyte procoagulant activ-

ity: requirement for antigen presentation and histocompati-bility leukocyte antigen-DR molecules. J Clin Invest 1985;76:

970-7.31. Taylor FB.Jr. Chang A. RufW. et al. Lethal E. coli septic shock

is prevented by blocking tissue factor wiili monoclonal antibody.

Circ Shock 1991;33:127-34.32. Edgington TS. Mackman N, Fan ST. RufW. Cellular immune and

cytokine pathwa}"S resulting in tissue factor expression and rele-

vance tO septic shock. Nouv Rev Fr HematoI1992;34(suppl):Sl5-

S27.33. Levy GA, Leibo\\1tz JL, Edgington TS. Induction of monocyte

procoagulant activity by murine hepatitis virus type 3 parallels

disease susceptibility in mice.J Exp Med 1981;154:1150-63.34. Williams JG. Maier RV. Ketoconazole inhibits alveolar macro-

phage production of intlammatory mediators involved in acute

lung injury (adult respiratory distress syndrome). Surgery 1992;

112:270-7.35. Helin H. Edgington TS. AIlogeneic induction of the human T

cell-instrllcted monocyte procoagulant response is rapid and

elicited by HLA-DRJ Exp Med 1983;158:962-75.36. Helin H. Edgington TS. A distinct "slow" cellular pathway

involving soluble mediators for the T cell-instrllcted induction of

monocyte tissue factoractivityin an allogeneic immune response.

J ImmunoI1984;132:2457-63.37. Rothberger H, Zimmerman TS. Spiegelberg HL. Vaughan

JH. Leukocyte procoagulant activity: enhancement of produc-tion in vitro by IgG and Ag:Ab complexes.J Clin Invest 1977;59:

549-57.38. Semeraro NA, Biondi A, Lorenzet R. Locati D.Mantovani A, Do-

nati MB. Direct induction oftissue factor by endotoxin in humanmacrophages from diverse anatomical sites. Immunology 1983;

50:529-35.39. ShandsJW. Lyrnphocyte cooperation is not required for induc-