Vol. 61, No. 10INFECrION AND IMMUNITY, Oct. 1993, p. 4293-43010019-9567/93/104293-09$02.00/0Copyright © 1993, American Society for Microbiology

Activation of the Complement System in Baboons Challengedwith Live Escherichia coli: Correlation with Mortality and

Evidence for a Biphasic Activation PatternJ. P. DE BOER,1 A. A. CREASEY,2 A. CHANG,3 D. ROEM,1 A. J. M. EERENBERG,1 C. E. HACK,`*

AND F. B. TAYLOR, JR.3Central Laboratory of the Netherlands Red Cross Blood Transfiusion Service and Laboratory for Experimentaland Clinical Immunology, University ofAmsterdam, Amsterdam, The Netherlands'; Chiron Corporation,

Emeryville, California 946082; and Oklahoma Medical Research Foundation, Oklahoma City,Oklahoma 731043

Received 5 March 1993/Returned for modification 12 May 1993/Accepted 26 July 1993

Activation of the complement system was studied in baboons that were challenged with live Escherichia coli.In the group challenged with a lethal dose (n = 4), the complement activation parameters C3b/c, C4b/c, andCSb-9 increased 13, 5, and 12 times the baseline value, respectively, during the first 6 h after the E. coliinfusion, whereas in the group challenged with a sublethal dose (n = 10), they increased only moderately, by2 to 3 times the baseline value. However, in this latter group, a more pronounced activation occurred at 24 h.Subsequent experiments showed that this second phase in complement activation started at 6 h after thechallenge, at which time infused microorganisms had been cleared from the circulation. The simultaneousincrease in C-reactive protein with this second phase suggested an endogenous activation mechanism involvingthis acute-phase protein. Levels of inactivated (modified) Cl inhibitor also increased in both groups, with peaklevels of 2.5 times the baseline value at 24 h in the sublethal group and of 4 times at 6 h after the challenge inthe lethal group. Thus, activation of complement in this animal model for sepsis occurs in a biphasic pattern,the initial phase mediated by the bacteria and the later phase mediated by an endogenous mechanism possiblyinvolving C-reactive protein. The differences in complement activation between animals with lethal or sublethalsepsis support the hypothesis that complement activation contributes to the lethal complications of sepsis.

The sepsis syndrome is caused by a host response tocirculating microorganisms and their products. This re-sponse comprises several inflammatory mediator systemsthat are either activated directly via components of theinvading microorganisms or indirectly, for example, via thegeneration of proinflammatory cytokines like tumor necrosisfactor alpha (TNF) and interleukin 11 (IL-1) (9). Excessiveactivation of these inflammatory mediator systems is sup-posed to play a pivotal role in the development of circulatoryshock and multiple organ failure and to contribute to the highmortality associated with sepsis (2, 21, 23).One of the mediator systems that is activated by invading

microorganisms is the complement system. For example,direct activation of this system by lipopolysaccharides (LPS)or even intact bacteria is achieved via the alternative path-way of complement (24). This activation results in thegeneration of the anaphylatoxins C3a and C5a, which havepowerful biological effects such as the induction of anincrease in vascular permeability, the contraction of smoothmuscles, the release of histamine from mast cells and baso-phils, and the recruitment of inflammatory cells by theirchemoattractant activities (1, 17, 40). In addition, C5a is apotent activator of neutrophils and enhances the LPS-in-duced synthesis of TNF and IL-1 by monocytes and macro-phages (27, 40). The membrane attack complex, i.e., theC5b-9 complex, is not only involved in the complement-mediated lysis of cells but also may induce activation ofcoagulation by induction of tissue factor activity on endo-thelial cells (5) and the formation of prothrombinase on

* Corresponding author.

platelets and endothelial cells (15, 29, 41), both of whichmechanisms may contribute to the development of multipleorgan failure (19). Thus, complement activation yields activefragments that in various ways may play a role in thepathogenesis of the sepsis syndrome.

Several studies have demonstrated activation of the com-plement system in sepsis, the extent of activation beingassociated with a lethal outcome (1, 3, 12, 17, 23). It is oftenassumed that this activation is directly triggered by micro-organisms or their products such as LPS. However, werecently observed a marked activation of complement inpatients with a septic shock-like syndrome induced by highdoses of IL-2 (37). In this condition, the observed activationwas not due to the presence of bacteria. Therefore, thisobservation points to an endogenous mechanism of comple-ment activation. To investigate whether in sepsis a similarendogenous activation mechanism occurs and to furtherdelineate the relation of complement activation to clinicalcourse and activation of other mediators, we decided toanalyze the changes in complement activation in baboonschallenged with a lethal or sublethal dose of Escherichia coli.Our results indicate that the complement system in theseseptic animals is activated in a biphasic pattern and that theextent of this activation is related to the clinical outcome.

MATERIALS AND METHODS

Baboon model for sepsis. The baboon model for sepsis thatwas used in this study has been described in detail elsewhere(6, 20, 34, 36). Briefly, a mixed breed of Papio C. cynoceph-alus and Papio C anubis baboons were purchased fromCharles River Breeding Laboratories, Inc. (Wilmington,

4293

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

4294 DE BOER ET AL.

Del.). The baboons were sedated with ketamine hydrochlo-ride (Aveco Company, Inc., Fort Dodge, Iowa; 14 mg/kg ofbody weight intramuscularly) on the morning of the studyand anesthetized with sodium pentobarbital (2 mg/kg). Theanimals were intubated orally and allowed to breathe spon-taneously. The femoral artery and both femoral veins werecannulated aseptically and used for measuring aortic pres-sure, obtaining blood samples, and the infusion of liveorganisms, fluids, and anesthetic. In these studies, either alethal dose (4 baboons [lethal group]), i.e., 4 x 1010 CFU/kg,or a sublethal dose (10 baboons [sublethal group]), i.e., 0.4 x1010 CFU/kg, of E. coli type B isolated as described previ-ously (20) was infused. The animals receiving the lethal dosedied between 6 and 10 h. The clinical and laboratory data ofthese 14 animals have been described elsewhere (39).

In a subsequent experiment, three additional baboonswere challenged with a sublethal dose of E. coli and onebaboon was challenged with a lethal dose of E. coli. Thelatter animal died 28 h after the challenge.Mean systemic arterial pressure and heart rate were

measured in each animal every hour during the 6-h observa-tion period. Hematologic parameters were assessed in bloodsamples collected at T+0, T+30, T+60, T+120, T+180,T+240, T+300, T+360, and T+1440 min as described previ-ously, where T is the time in minutes after the start of thechallenge (34). In addition, at each of these time points, 5-mlblood samples were also collected in 10 mM EDTA and0.05% (wt/vol) Polybrene (final concentrations) as describedpreviously (8). The plasma samples were stored in aliquots at-70°C. The study protocol used received prior approval bythe Institutional Animal Care and Use Committees of theOklahoma Medical Research Foundation and the OklahomaHealth Sciences Centers.

Reagents. CNBr-activated Sepharose 4B was purchasedfrom Pharmacia Fine Chemicals (Uppsala, Sweden), hexa-dimethrine bromide (Polybrene) was from Janssen Chimica(Beerse, Belgium), and Tween 20 was from J. T. BakerChemical Co. (Phillipsburg, N.J.). Purified human neutrophilelastase was purchased from Elastine Products Co., Inc.(Pacific, Mo.). Cobra venom factor (CVF) was purchasedfrom Feinbiochemica (Heidelberg, Germany).

Assays. All assays used in this study were developed withmono- or polyclonal antibodies raised against human pro-teins. However, some of the antibodies reacted only poorlywith the baboon proteins. Therefore, assays in which theselatter antibodies were used had to be modified for the presentstudy (vide infra). When possible, standards were preparedwith baboon plasma or serum to correct for differences inaffinities of the antibodies for the baboon versus the humanproteins. When human standards had to be used, we estab-lished that dilution curves of these standards were parallel tothose obtained with baboon plasma samples.Normal values in baboons were calculated for all assays

by using the values in plasma samples obtained from theanimals before the challenge.

(i) RIA for iCl-INH. Proteolytically inactivated Cl inhib-itor (iCl-INH) was assessed in baboon plasma with a radio-immunoassay (RIA) as described previously (26). In brief,monoclonal antibody (MAb) KII that specifically binds iCl-INH was coupled to Sepharose and incubated by head-over-head rotation with the samples to be tested. iCl-INH boundto the Sepharose beads was quantitated by a subsequentincubation with 125I-anti-Cl inhibitor (1251I-anti-Cl-INH) an-tibodies. The results were expressed as a percentage of theamount of iCl-INH present in normal baboon plasma (NBP).

(ii) RIA for C1-C1-INH complexes. C1-C1-INH complexes

were assessed in baboon plasma with an RIA in which MAbKOK12 (against complexed C1-INH) was used as captureantibody and polyclonal 1251I-anti-Cl antibodies were used asdetecting antibodies (26). Results were expressed as a per-centage of the amount of complexes present in normalbaboon serum (NBS) in which a maximal amount of Cl-Cl-INH complexes was generated by incubation of 1 volume ofNBS with 1 volume of heat-aggregated human immunoglob-ulin G (AHG; combination termed NBS-AHG), exactly asdescribed previously for human serum (11).

(iii) RIA for C3b/c. C3b/c in baboon plasma was assessedwith an RIA by using MAb anti-C3-28, which binds to aneoepitope expressed on C3b, C3bi, and C3c after disruptionof the thiolester, and polyclonal 1251I-anti-C3c antibodies (13,14). Results were expressed as a percentage of the amount ofC3b/c present in normal baboon serum aged (NBA), i.e.,NBS incubated for 7 days at 37°C in the presence of NaN3(0.02% [wt/vol]), as described previously for human serum(13). In experiments in which human C3b/c was assessed,normal human serum (NHS) aged, prepared as described forNBA, was used as a standard.

(iv) RIA for C4b/c. C4b/c in baboon plasma was assessedwith an RIA similar to that described for the determinationof C3b/c. However, in this RIA, MAb anti-C4-1, which bindsto a neoepitope expressed on human C4b, C4bi, and C4cafter disruption of the thiolester, and polyclonal 1251I-anti-C4antibodies were used (42). Results were expressed as apercentage of the amount of C4b/c present in NBS-AHG.C4b/c in human serum was expressed as a percentage of theamount of C4b/c present in NHS-AHG (11).

(v) Quantification of total C3 and C4. The antigenic levelsof C3 and C4 were measured with a Behring nephelometricanalyzer (Behringwerke AG, Marburg, Germany), as de-scribed in the manufacturer's instructions and by usingantisera raised against human proteins. Results were ex-pressed in international units per milliliter by reference to ahuman standard plasma (HOO-4-NZT; Department of Im-mune Reagents, Central Laboratory of the Netherlands RedCross Blood Transfusion Service) that contained 108 IU ofC3 per ml and 99 IU of C4 per ml.

(vi) RIA for elastase-ad-antitrypsin complexes. Levels ofelastase-al-antitrypsin complexes were assessed with anRIA as described previously (25). Results were expressed innanomoles per liter by reference to a standard curve of NBPin which 66.6 nmol of elastase-al-antitrypsin complexes perliter was generated by incubating 1 volume of human neu-trophil elastase (10 ,ug/ml in phosphate-buffered saline[PBS]) with 4 volumes of NBP for 15 min at room tempera-ture.

(vii) RIA for PAP complexes. Plasmin-a2-antiplasmin(PAP) complexes were determined with an RIA as describedpreviously (22). Slight modifications were made to allowmeasurements in baboon plasma (8). Levels of PAP com-plexes were expressed as a percentage of the level present inNBP in which a maximal amount of PAP complexes wasgenerated by incubation with urokinase (8).

(viii) ELISA for C5b-9 complexes. Levels of the terminalcomplex of complement (C5b-9) in baboon plasma wereassessed with an enzyme-linked immunosorbent assay(ELISA), kindly provided by Behringwerke, as described inthe instructions of the manufacturer. Results were expressedin micrograms per liter by reference to a standard of humanproteins provided by the manufacturer.

(ix) ELISA for CRP. Plasma levels of C-reactive protein(CRP) were assessed in an ELISA in which monospecificrabbit anti-human CRP antibodies (obtained from the De-

INFEcr. IMMUN.

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

COMPLEMENT ACTIVATION IN E. COLI-CHALLENGED BABOONS 4295

TABLE 1. Actual baseline values of the parameters measureda

Parameter SLECb LDECC

iCl-INH (% NBP) 105.35 ± 45.45 86.62 ± 45.06C1-C1-INH (% NBS-AHG) 1.18 ± 0.41 1.83 + 0.99C3b/c (% NBA) 0.15 ± 0.07 0.11 + 0.05C4b/c (% NBS-AHG) 1.17 ± 0.28 1.35 ± 0.48C5b-9 (i±g/liter) 119.10 + 32.56 156.50 ± 28.57Elastase-al-antitrypsin 0.12 ± 0.02 0.10 ± 0.03

(nmol/iter)PAP (% NBP-MA-UK)d 0.61 ± 0.26 0.58 + 0.09CRP (AU/ml)r 0.99 ± 0.75 1.07 ± 0.52

a Mean ± SD.b SLEC, baboons that received the sublethal dose of E. coli (n = 10).c LDEC, baboons that received the lethal dose of E. coli (n = 4).d NBP-MA-UK, NBP in which a maximal amount of PAP complexes was

generated by addition of urokinase (8).' AU, arbitrary unit.

partment of Immune Reagents) were used as capture anddetecting antibodies. In brief, anti-CRP antibodies werecoated overnight at room temperature (5 jg/ml in PBS, 100,ul per well) on microtiter plates (Maxisorp; Nunc, Roskilde,Denmark). After a washing procedure, serial dilutions ofbaboon plasma (100 ,ul per well) were incubated with bioti-nylated anti-CRP antibodies (1.8 p,g/ml) for 2 h at roomtemperature. The plates were then washed and incubatedwith streptavidin horseradish peroxidase diluted 1:1,000 inPBS-0.1% (wt/vol) Tween 20-0.2% (wt/vol) gelatin. After awashing procedure, the plates were developed with a solu-tion of 3,5,3',5'-tetramethylbenzidine with 0.0003% H202 in0.11 M sodium acetate (pH 5.5; 100 p,l per well). Thereafter,the reaction was stopped by the addition of an equal volumeof 2 M H2SO4. The plates were read at 450 nm in a TitertekMultiscan reader. Results were related to a standard of NBP,which was defined to contain 1 arbitrary unit of CRP per ml.

Statistical analysis. Three (at T+180), 1 (at T+240), and 2(at T+360) samples in the lethal group of animals and 6 (atT+240), 8 (at T+360), and 7 (at T+1440) samples in thesublethal group were available for the present investigation,whereas at the other time points, the indicated 10 and 4samples were available in the sublethal and the lethal groups,respectively. The parameters which were assessed in thebaboons were expressed as means and standard deviations(SD) of changes relative to the baseline value (i.e., at T+0) ofeach animal. The mean + SD of the baseline values in actualunits are presented for both groups of baboons in Table 1.Statistical analysis was performed on an IBM-compatiblecomputer with an SSPS statistical package. Differencesbetween both groups of animals were analyzed by using theMann-Whitney-Wilcoxon rank sum test. Differences in thesublethal group between various time points were analyzedwith Student's t test. A difference was considered signifi-cant, by using a two-tailed P, at P < 0.05 and highlysignificant at P < 0.005. Spearman's rank correlation analy-sis was used to determine the correlation between C5b-9 andCRP levels in the animals that were sampled for more than 6h after the E. coli challenge.

RESULTS

Validation of the assays for C3b/c and C4b/c in baboons. Wehave recently developed MAbs that allow the determinationof C4b/c and C3b/c in human plasma (14, 42). These MAbscould also be used to specifically detect baboon C4b/c andC3b/c as was demonstrated in experiments that tested sam-

100

z

IL0

:0

C3b/c

76

60 F

26

100

a

cInzIL0

W

a

C4b/c

76

60

25

,ftvCVF EDTA+CVF AHQ ETA+Al Ms

FIG. 1. Validation of the assays of C3b/c (A) and C4b/c (B) inbaboons. One volume ofNBS was incubated with either 1 volume ofAHG (3 mg/ml), 1 volume of CVF (20 U/ml), or 1 volume of saline(PBS) for 30 min at 37°C. Thereafter, the reaction was stopped bythe addition of 0.5 volume of ice-cold 0.5 M EDTA (pH 7.4), andmixtures were analyzed for C3b/c and C4b/c contents as describedin Materials and Methods. In control experiments, 1 volume of NBSwas incubated with 0.5 volume of 0.5 M EDTA (pH 7.4) for 15 minat 0°C prior to incubation with AHG or CVF. C3b/c and C4b/c are

expressed as percentages of the amount of C3b/c and C4b/c presentin NBA and NBS-AHG, respectively.

ples of NBS that were incubated either for one-half hour at37°C with an equal volume of AHG (3 mg/ml in PBS) toactivate the classical pathway or with an equal volume ofCVF (20 U/ml in PBS) to activate the alternative pathway.Incubation of NBS with AHG yielded both a measurableincrease of C4b/c to 98.3% of the NBS-AHG standard aswell as an increase of C3b/c to 50.8% of the NBA standard(Fig. 1). As expected, generation of only C3b/c, i.e., to43.7% of the NBA standard, was observed when NBS wasincubated with CVF. Preincubation of NBS with 0.5 MEDTA (pH 7.4) for 15 min at 0°C prior to adding bothactivators inhibited the increase of both complement activa-tion parameters (Fig. 1). This experiment was done twiceand yielded results identical to those observed previously forhuman serum (42). Therefore, we concluded that the assayscould be used to specifically detect baboon C4b/c and C3b/c.

Activation parameters of the complement system in baboonsafter lethal and sublethal E. coli challenge. Administration ofa lethal dose of E. coli resulted in a pronounced activation ofthe complement system. As shown in Fig. 2A, almostimmediately after the start of the E. coli infusion, levels ofC3b/c steeply increased to eight times the baseline value atT+60. After a slight decrease, a peak value of 13 times thebaseline value occurred at T+ 180.

Circulating C4b/c also sharply increased to five times thebaseline value at T+ 180 (Fig. 2B), indicating that at least

1B

lA

VOL. 61, 1993

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

4296 DE BOER ET AL.

-J

a]0,

lU0

w

0)

ulz

Fa:

cc

w

lU

w

0)

max

w

IL

-.

Lu

0)

tL

0,

20

10

0

9

7

5

3

12

20

10

0

0 100 200 300 400

TIME IN MINUTESFIG. 2. Levels of the complement activation products (mean and

SD, expressed as changes relative to baseline values) C3b/c (A),C4b/c (B) and C5b-9 (C) in baboons who received a lethal (opencircles, n = 4) or sublethal (open triangles, n = 10) dose of E. coliintravenously. Significant differences between both groups at eachtime point are indicated with * (P < 0.05) or ** (P < 0.005).

part of the C3 was activated via the classical pathway.Plasma levels of the membrane attack complex, i.e., C5b-9,also rose to 12 times the baseline value at T+60 (Fig. 2C).These high levels of C5b-9 complexes remained constantuntil T+240, after which a slight decrease to eight times thebaseline value at T+360 was seen.

In contrast with the marked and rapid increase in circu-lating levels of complement activation products observed in

TABLE 2. Complement activation at 3 (T+ 180) and 24 (T+ 1440)h after administration of live E. coli

Complement activation atA:Complement Baboon groupa

T+180 T+ 1440

C3b/c SLEC 1.76 ± 0.85* 5.95 ± 3.64*LDEC 12.89 ± 3.76 NDC

C4b/c SLEC 1.33 ± 0.36* 7.65 ± 6.62*LDEC 5.48 ± 3.68 ND

C5b-9 SLEC 1.71 ± 0.43** 4.59 ± 1.93**LDEC 11.37 ± 1.02 ND

a For definitions of SLEC and LDEC, see Table 1, footnotes b and c. Forthe lethal group, n equals 3; for the sublethal group, n equals 10 at T+180 andn equals 7 at T+1440.

b Expressed as increases relative to baseline values (mean ± SD). * orP < 0.05 or P < 0.005, respectively, for the difference between values atT+180 and T+1440.

c ND, not determined.

the animals that received the lethal dose of E. coli, only amodest increase in these parameters was observed in ba-boons challenged with the sublethal dose. During the first 6h, a small increase with peak values of three times thebaseline value at T+120 in C3b/c levels occurred (Fig. 2A),whereas during this period no significant changes in C4b/clevels were noted (Fig. 2B). Finally, a moderate increase oftwo times the baseline value was observed for plasma levelsof C5b-9 complexes during the first 6 h (Fig. 2C). Nosignificant changes were observed in total C3 and C4 levelsin both groups during the observation period (data notshown).The rapid rise in circulating complement activation prod-

ucts upon infusion with the lethal dose of E. coli suggesteddirect activation of the complement system by E. coli. Fromthe animals challenged with the sublethal dose of E. coli,samples obtained at 24 h after the E. coli administration werealso available for analysis. Surprisingly, levels of all com-plement activation parameters measured were increased atthis time point in all animals tested (n = 7) and exceeded themaximal levels observed during the first 6 h (Table 2). Thissuggested the presence of an activation mechanism otherthan that triggered directly by the bacteria or their constitu-ents. However, samples taken from the animals during theperiod from 6 to 24 h after the challenge were not available.Therefore, we decided to challenge three other baboons witha sublethal dose and one with a lethal dose of E. coli tofurther substantiate this presumed biphasic activation pat-tern of complement. In the three baboons challenged with asublethal dose, an initial rise in C5b-9 complexes wasobserved, with peak levels at T+ 120 (Fig. 3). A second risein C5b-9 complexes, which started at T+360 and exceededthe activation seen during the first few hours, was thenobserved (Fig. 3). In the baboon challenged with a lethaldose, the second rise in C5b-9 complexes was less pro-nounced than the initial activation observed during andimmediately after the E. coli infusion. Nevertheless, levelsat T+ 1440 represented a fivefold increase compared with thebaseline value (the four previous animals that received thelethal dose of E. coli had died at this time point). The secondincrease in C5b-9 complexes in the three baboons challengedwith a sublethal dose was accompanied by a steep rise inplasma levels of CRP (Fig. 3A), which suggested the involve-ment of this acute-phase reactant in the second phase ofcomplement activation. Spearman rank analysis of the C5b-9and CRP levels from T+240 to T+1440 yielded a correlation

INFECT. IMMUN.

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

COMPLEMENT ACTIVATION IN E. COLI-CHALLENGED BABOONS 4297

0 6 12 18 24

TIME IN HOURS

FIG. 3. Levels of C5b-9 complexes (solid lines), expressed as

changes relative to the baseline value, and CRP (broken lines),expressed as arbitrary units per milliliter, in four baboons whoreceived either a sublethal (A; n = 3, data represent mean and SD)or a lethal (B; n = 1) dose of E. coli and were sampled for more than6 h after the challenge.

coefficient of 0.64 (P < 0.001). In the baboon challenged witha lethal dose, a marked increase in CRP was also observed,but probably because of the excessive complement activa-tion in the initial phase and the low number of samples, noclear correlation was seen between the rise in CRP andelevated C5b-9 levels during this period (Fig. 3B).

Proteolytic inactivation of Cl-INH in baboons after E. colichallenge. In a former study, we showed increased levels ofiCl-INH in patients with sepsis. The levels of iCl-INH inpatients on admission to an intensive care unit correlatedwith mortality (26). Therefore, we also measured iCl-INHlevels in the septic baboons. As shown in Fig. 4, circulatingiCl-INH in the lethal group increased immediately after thestart of the E. coli administration, yielding levels at T+360that were four times the baseline value. In contrast with theincrease in the lethal group, there was no change in iCl-INHin the sublethal group until T+240. Thereafter, a slightincrease was observed to two times the baseline values atT+360, whereas the highest iCl-INH levels occurred atT+1440 (Table 3).

Possible proteinases that may have inactivated C1-INHare activated Cl, plasmin, and neutrophilic elastase (26). Toelucidate the mechanism by which the inactivation of Cl-INH in the baboons could have occurred, levels of Cl-Cl-INH, elastase-al-antitrypsin, and PAP complexes were as-sessed as parameters that reflected activation of Cl, the

660

40

;30

z

0

D

L0

UJ0

w<)

lU

20

10

5

4

2

1'1

500

0 100 200 300 40040

TIME IN MINUTESFIG. 4. iCl-INH (mean and SD, expressed as changes relative to

30 baseline values) in baboons who received a lethal (open circles, n =4) or sublethal (open triangles, n = 10) dose ofE. coli intravenously.Significant differences at each time point between both groups are

20 indicated with * (P < 0.05).

release of elastase by neutrophils, and the generation ofplasmin in vivo, respectively. In Table 3, these parametersare compared with iCl-INH levels at various time points.Both the increase in C1-C1-INH as well as that in elastase-al-antitrypsin complexes in the sublethal group paralleledthe increase of iC1-INH at T+360 and T+1440. AlthoughPAP complexes in the sublethal group were also increased atthese time points, the highest levels of PAP did not coincidewith those of iC1-INH. In the lethal group, increased levelsof iC1-INH at T+120 and T+360 were accompanied byincreases of elastase-al-antitrypsin complexes to 45 and 40times the baseline value, respectively, and by increases ofPAP levels to 19 and 11 times the baseline value. In contrast,C1-C1-INH complexes were only modestly increased atthese time points (Table 3).

Activation of neutrophils via activation of the complementsystem. The administration of a lethal or sublethal dose of E.coli to baboons resulted in a sharp fall in circulating leuko-cytes during the first 2 h which is probably due to attachmentto the endothelium (8). Damage of the endothelium duringsepsis is in part mediated via activation of attached neutro-phils (31). Since C5a is a powerful inducer of neutrophilactivation, we investigated the relation of complement acti-vation to the activation of neutrophils. In the lethal group,elastase-al-antitrypsin complexes increased to a peak levelof 45 times the baseline value at T+120. After a slight fall atT+ 180, a second peak of 50 times the baseline value atT+240 was then observed (Fig. 5). The elastase-al-antitryp-sin levels in the sublethal group showed a more gradualincrease to a peak level of 18 times the baseline value atT+360 and were still elevated by 20 times the prechallengelevels at T+1440 (Table 3).

In vitro activation of the complement system by live E. coli.The fast onset of complement activation after E. coli chal-lenge suggested that the initial activation of complement wasdirectly mediated by the bacteria. Therefore, we investi-gated in vitro whether E. coli could activate the complementsystem in baboon serum. For comparison, similar experi-ments were done with human serum. One volume of live E.coli was incubated for one-half hour at 37°C with either 1

IC1 -INH *

I

iI-T T

VOL. 61, 1993

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

4298 DE BOER ET AL.

TABLE 3. Comparison of levels of iCl-INH with levels of target and nontarget proteases for C1-INH at various time points

Increase (mean + SD) at":Inhibitor or complex Baboon groupa

T+120 T+360 T+1440

iCl-INH SLEC 0.92 ± 0.17* 1.73 + 1.34 2.45 ± 1.26LDEC 2.70 + 1.67* 3.72 + 0.67 NDC

C1-C1-INH SLEC 1.08 ± 0.37 1.55 ± 0.33 2.13 ± 0.62LDEC 1.53 + 0.55 1.64 ± 0.51 ND

Elastase-cal-antitrypsin SLEC 9.92 ± 4.94** 18.16 ± 5.37** 20.25 ± 6.9LDEC 44.78 3.37** 40.50 ± 0.71** ND

PAP SLEC 2.60 ± 2.12** 1.33 ± 0.62** 1.20 ± 0.26LDEC 19.04 ± 10.15** 11.29 ± 5.11** ND

a For definitions of SLEC and LDEC, see Table 1, footnotes b and c. For the sublethal group, n equals 10, 8, and 7 at T+ 120, T+360, and T+ 1440, respectively;for the lethal group, n equals 4 and 2 at T+120 and T+360, respectively.

I All parameters expressed as increases relative to baseline values. * or * P < 0.05 orP < 0.005, respectively, for the differences between the sublethal andlethal groups at the indicated time points.

c ND = not determined.

volume of NBS or 1 volume of NHSreaction was stopped by the subsequent irvolume of 0.5 M EDTA (pH 7.4) for 15 mirin Fig. 6, a dose-dependent increase ofactivation was observed upon incubation ocoli. Assuming a plasma volume of 40 ml/lnumber of circulating organisms in the aniithe lethal dose would be approximately 11the infusion of E. coli. In vitro, this nuncaused a significant activation of both thealternative pathways (Fig. 6).

DISCUSSION

This study demonstrates that administdose of E. coli to baboons results in a pronof the complement system. The steep inment activation parameters in the lethal gafter the onset of the E. coli challenge s

w

z

ax0)

IL0

6cc

a:

60

60

40

30

20

10

0 100 200

TIME IN MINUTESFIG. 5. Elastase-oal-antitrypsin complexes I

pressed as changes relative to baseline valuereceived a lethal (open circles, n = 4) or subleth= 10) dose of E. coli intravenously. Significanttime point between both groups are indicated w** (P < 0.05).

,. Thereafter, the initial activation of the complement system was directlyicubation with 0.5 mediated by circulating E. coli. Indeed, in vitro studies havei at 0°C. As shown shown that endotoxin and even intact bacteria may directlyC3b/c and C4b/c activate the complement system in particular via the alter-

)fNBS with live E. native pathway (24). However, the initial complement acti-kg in baboons, the vation observed in the baboons had occurred at least in partmals that received via the classical pathway since C4b/c levels also increased09/ml shortly after upon challenge with E. coli. Further experiments showednber of organisms that the E. coli organisms generated both C3b/c as well asclassical and the C4b/c in NBS (Fig. 6) when tested at concentrations similar

to those which occur shortly after infusion (i.e., 108 to 109organisms per ml). In addition, the dose-dependent activa-tion of complement by the E. coli organisms in these in vitroexperiments suggested that the severity of the initial activa-

tration of a lethal tion of the complement system in this animal model wasLounced activation determined by the number of circulating E. coli organisms.crease in comple- Consistent herewith, the initial activation of the complementJroup immediately system preceded the appearance of TNF-a, IL-11, and IL-6;uggested that the (6), suggesting that this initial activation occurred indepen-

dently from processes which are induced by these proinflam-matory cytokines. We assumed that the observed activationof the classical pathway was due to the presence of immu-noglobulin G and/or M antibodies against the E. coli bacteriabut considered it beyond the scope of this study to furtherevaluate the mechanism of the in vitro activation of comple-

* ment by the E. coli organisms.The administration of a sublethal dose ofE. coli resulted in

a moderate activation of complement during the first hoursof the experiments. Surprisingly, all samples obtained fromthe animals at 24 h (n = 7) contained higher levels ofcomplement activation products than were present in any ofthe samples taken during the first 6 h after the challenge(Table 2), suggesting a second phase of activation. Addi-tional experiments in three animals that received a sublethaldose of E. coli revealed that this second phase of comple-ment activation started at 4 to 6 h after the start of the E. colichallenge (Fig. 3). Apparently, this second phase of comple-ment activation was not due to direct activation of the

300 400 system by intact E. coli or endotoxin since neither isdetected in the circulation after 4 to 6 h (33). More likely, thissecond response was due to an endogenous activation mech-

(mean and SD, ex- anism.s) in baboons who The observation of a second phase of complement activa-ial (open triangles, n tion in the septic baboons raises several questions. First, thedifferences at each mechanism of this activation is not clear. One possibility is

iith * (P < 0.05) and that it is mediated by cell constituents exposed after tissuedamage. For example, cellular compounds can activate

INFECT. IMMUN.

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

COMPLEMENT ACTIVATION IN E. COLI-CHALLENGED BABOONS 4299

100

100

126

|t| ~~~~~~~C4b/c

IOEIO IOE9 1OE8 lOE7 EDTA P8S

FIG. 6. In vitro activation of the complement system, reflectedby C3b/c and C4b/c levels, by live E. coli in baboons (solid bars) andhumans (dotted bars). One volume of either NBS or NHS wasincubated for 30 min at 37°C with 1 volume of E. coli (i.e., 1010, 109,108, and 107 CFU of live E. coli per ml [lOE10, 10E9, 10E8, and10E7, respectively]) or saline (PBS). The reaction was stopped by asubsequent incubation for 15 min at 0°C with 0.5 volume of 0.5 MEDTA (pH 7.4), and mixtures were analyzed for C3b/c and C4b/ccontents as described in Materials and Methods. As a control, 1volume of NBS or NHS was also incubated with 0.5 volume of 0.5M EDTA (pH 7.4) prior to incubation with the microorganisms andsubsequently incubated with 1 volume of 1010 CFU of live E. coli perml. C3b/c levels are expressed as percentages of the amount ofC3b/c present in NBA (baboons) and NHS aged (humans), andC4b/c levels are expressed as percentages of the amount of C4b/cpresent in NBS-AHG or NHS-AHG.

complement as has been demonstrated in myocardial infarc-tion (10, 28). In favor of this hypothesis is that the secondphase of complement activation occurred at the stages oftissue damage and degeneration described previously in thisbaboon model (33).

Alternatively, the second phase of activation could be dueto a mechanism induced by cytokines. In a previous study(6), we have shown in another series of baboons that wereadministered live E. coli that levels of TNF-a started toincrease at 60 min after the challenge, reaching peak valuesat 120 min. This peak of TNF-a was followed by a peak ofIL-10 at 5 h and by an increase of IL-6, which in the lethalanimals continued beyond 6 h. We did not measure circulat-ing levels of these cytokines in all animals described in thisstudy (because of the limited volumes of the plasma sam-

ples). However, in four animals (two with a sublethal doseand two with a lethal dose), we measured TNF-a and IL-6,and the course of these cytokines was similar to thatobserved in the previous study (6). Therefore, the release ofproinflammatory cytokines preceded the second phase ofcomplement activation, suggesting that this second phasewas induced by cytokines. High doses of IL-2 cause activa-

tion of complement in vivo (37). Recently, Vachino et al.suggested the involvement of CRP in this IL-2-inducedactivation of complement (38). Our observation that theincrease of CRP levels coincided with the second phase ofcomplement activation in the septic baboons (Fig. 3)strongly suggested that CRP was involved in mediating thissecond phase of complement activation in the septic ba-boons. IL-6 is the cytokine that in vivo induces CRPresponses (18). Moreover, inhibition of IL-6 in vivo attenu-ates sepsis in mice (30). Further studies are needed toestablish whether the anti-IL-6 effects are explained byinhibition of CRP-dependent complement activation.Another question is how does this second phase of acti-

vation contribute to the clinical course? The animals thatreceived the lethal dose of E. coli died within 6 to 10 h.Unfortunately, we did not collect samples later than 6 h afterthe challenge. Therefore, it is unclear whether the death inthese animals coincided or even was due to the second phaseof complement activation. In the animals challenged with asublethal dose, a more pronounced complement activation at24 h was accompanied by a further increase in neutrophilactivation as reflected by rising levels of elastase-al-anti-trypsin complexes (Table 3). At the same time, a deteriora-tion of the function of several organs was noted in theseanimals; for example, the plasma levels of liver transami-nases and creatinine increased (data not shown). One mayspeculate, therefore, that the second phase of complementactivation contributed to the development of multiple organfailure in these animals via the activation of neutrophils. Athird question regarding the biphasic activation of comple-ment is to what extent the second phase, possibly CRPmediated, contributes to the activation observed in humansepsis. Obviously, this issue needs further investigation.

Proteolytic inactivation of C1-INH has been shown tooccur in severely septic patients, and levels of iCl-INH inpatients with sepsis on admission have prognostic signifi-cance (26). Our observation that levels of iCl-INH were thehighest in the lethal group (Fig. 4) are in agreement withthose findings. Inactivation of C1-INH during sepsis can bemediated by several pathways (26), including inappropriatecomplex formation of this serine proteinase inhibitor with itstarget proteinases such as Cl (41) and proteolytic inactiva-tion by nontarget proteinases like elastase and plasmin (4).The peak values of iC1-INH in the sublethal group coincidedon all occasions with peak values of elastase-al-antitrypsinand C1-C1-INH complexes but not with those of PAPcomplexes (Table 3). This virtually eliminates the possibilitythat iCl-INH was generated by interaction with plasmin butleaves open the possibility that either Cl or elastase or bothwere responsible for the generation of iCl-INH.The early and pronounced activation of the complement

system in baboons challenged with a lethal dose of E. colisuggested that the complement system is involved in thelethal outcome. Support for this notion is provided by animalstudies that have demonstrated a protective effect of anti-C5a antibodies on the development of shock and multipleorgan failure in sepsis (16, 32). The precise mechanism(s)that mediates the lethal effects of complement activation isnot known. The simultaneous increases in C5b-9 and elas-tase-al-antitrypsin complexes in the lethal group of baboonssupport that complement products activate neutrophils dur-ing the initial stages of sepsis. In addition, indirect effects ofcomplement on neutrophils might further contribute to thedevelopment of a lethal outcome of sepsis. For example, C5aenhances in vitro the endotoxin-mediated release of proin-flammatory cytokines by mononuclear cells (27), and these

VOL. 61, 1993

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

4300 DE BOER ET AL.

cytokines in turn prime neutrophils for C5a-mediated acti-vation (7). Previous studies have shown a marked activationof coagulation in this baboon model and a reduction inmortality by the administration of some anticoagulants (8,33-36). Recent in vitro studies indicate that activation of thecomplement system, i.e., generation of C5b-9 complexes,may contribute to activation of coagulation by the expres-sion of tissue factor on endothelial cells and the release ofmembrane vesicles from endothelial cells and plateletswhich contain prothrombinase activity (5, 15, 29). Whetherthese mechanisms are operating in vivo during sepsis isunknown, but the observation that complement is one of thefirst mediators to become activated in animals with lethalsepsis, together with the notion that in these animals thecoagulation system is activated before cytokines are de-tected (6, 8), is consistent with such a mechanism. Appar-ently, intervention studies with complement inhibitors areneeded to resolve this issue.

In conclusion, we demonstrated a biphasic activation ofcomplement in baboons challenged with live E. coli. Thisactivation was more pronounced in the animals that receivedthe lethal dose. These data are consistent with the notionthat complement activation is involved in the pathogenesisof the lethal complications of sepsis.

ACKNOWLEDGMENTS

We thank Heleen van Manen for her help in preparing themanuscript and Wim Schaasberg for his assistance in the statisticalanalysis of our data.This study was financially supported in part by grant 900-512-121

from the Netherlands Organization for Scientific Research (NWO).

REFERENCES

1. Bengtson, A., and M. Heideman. 1988. Anaphylatoxin formationin sepsis. Arch. Surg. 123:645-649.

2. Bone, R. C. 1991. The pathogenesis of sepsis. Ann. Intern. Med.115:457-469.

3. Brandtzaeg, P., T. E. Mollness, and P. Kierulf. 1989. Comple-ment activation and endotoxin levels in systemic meningococcaldisease. J. Infect. Dis. 160:585-65.

4. Brower, M. S., and P. C. Harpel. 1982. Proteolytic cleavage andinactivation of a2-plasmin inhibitor and Cl-inactivator by hu-man polymorphonuclear leucocyte elastase. J. Biol. Chem.257:9849-9854.

5. Carson, S. D., and D. R. Johnson. 1990. Consecutive enzymecascades: complement activation at the cell surface triggersincreased tissue factor activity. Blood 762:361-367.

6. Creasey, A. A., P. Stevens, J. Kenney, A. C. Allison, K. Warren,R. Catlett, L. Hinshaw, and F. B. Taylor, Jr. 1991. Endotoxinand cytokine profile in plasma of baboons challenged with lethaland sublethal Escherichia coli. Circ. Shock 33:84-91.

7. Crouch, S., and J. Fletcher. 1991. The priming effects of theproducts of stimulated mononuclear cells on the response ofneutrophils to C5adesarg. Br. J. Haematol. 77:158-164.

8. de Boer, J. P., A. A. Creasey, A. Chang, D. Roem, M. C.Brouwer, A. J. M. Eerenberg, C. E. Hack, and F. B. Taylor, Jr.1993. Activation patterns of coagulation and fibrinolysis inbaboons following infusion with lethal or sublethal dose of E.coli. Circ. Shock 39:59-67.

9. Dinarello, C. A. 1991. The proinflammatory cytokines interleu-kin-1 and tumor necrosis factor and treatment of the septicshock syndrome. J. Infect. Dis. 163:1177-1184.

10. Entman, M. L., L. Michael, R. D. Rossen, W. J. Dreyer, D. C.Anderson, A. A. Taylor, and C. W. Smith. 1991. Inflammation inthe course of early myocardial ischemia. FASEB J. 5:2529-2537.

11. Hack, C. E., A. J. Hannema, A. J. M. Eerenberg, T. A. Out, andR. C. Aalberse. 1981. A Cl-inhibitor-complex assay (INCA): a

method to detect Cl activation in vitro and in vivo. J. Immunol.127:1450-1453.

12. Hack, C. E., J. H. Nujens, R. J. F. Felt-Bersma, W. 0.Schreuder, A. J. M. Eerenberg-Belmer, J. Paardekooper, W.Bronsveld, and L. G. Thus. 1989. Elevated plasma levels of theanaphylatoxins C3a and C4a are associated with a fatal outcomein sepsis. Am. J. Med. 86:20-26.

13. Hack, C. E., J. Paardekooper, R. J. T. Smeenk, J. J. Abbink,A. J. M. Eerenberg, and J. H. Nuiens. 1988. Disruption of theinternal thioester bond in the third component of complement(C3) results in the exposure of neodeterminants also present onactivation products of C3. An analysis with monoclonal anti-bodies. J. Immunol. 141:1602-1609.

14. Hack, C. E., J. Paardekooper, and F. van Milligen. 1990.Demonstration in human plasma of a form of C3 that has theconformation of "C3b-like C3". J. Immunol. 144:4249-4255.

15. Hamilton, K. K., R. Hattori, C. T. Esmon, and P. J. Sims. 1990.Complement proteins C5b-9 induce vesiculation of the endothe-lial plasma membrane and expose catalytic surface for assemblyof the prothrombinase enzyme complex. J. Biol. Chem. 265:3809-3814.

16. Hangen, D. H., J. H. Stevens, P. S. Satoh, E. W. Hall, P. T.O'Hanley, and T. A. Raffin. 1989. Complement levels in septicprimates treated with anti-C5a antibodies. J. Surg. Res. 46:195-199.

17. Heideman, M., B. Norder-Hansson, A. Bengtson, and T. E.Mollnes. 1988. Terminal complement complexes and anaphyla-toxins in septic and ischemic patients. Arch. Surg. 123:188-192.

18. Heinnich, P. C., J. V. Castell, and T. Andus. 1990. Interleukin-6and the acute phase response. Biochem. J. 265:621-636.

19. Hesselvik, J. F., M. Blomback, B. Brodin, and R. Maller. 1989.Coagulation, fibrinolysis, and kallikrein systems in sepsis: rela-tion to outcome. Crit. Care. Med. 17:724-733.

20. Hinshaw, L. B., D. J. Brackett, L. T. Archer, B. K. Beller, andM. F. Wilson. 1983. Detection of the "hyperdynamic state" ofsepsis in the baboon during lethal E. coli infusion. J. Trauma23:361-368.

21. Kalter, E. S., M. R. Daha, J. W. Ten Cate, J. Verhoef, and B. N.Bouma. 1985. Activation and inhibition of Hageman factor-dependent pathways and the complement system in uncompli-cated bacteremia or bacterial shock. J. Infect. Dis. 151:1019-1027.

22. Levi, M., J. P. de Boer, D. Roem, J. W. Ten Cate, and C. E.Hack. 1992. Plasminogen activation in vivo upon intravenousinfusion of DDAVP: quantitative assessment of plasmin-a2-antiplasmin complexes with a novel monoclonal antibody basedradioimmunoassay. Thromb. Haemostasis 67:111-116.

23. McCabe, W. R. 1973. Serum complement levels in bacteremiadue to gram-negative organisms. N. Engl. J. Med. 288:21-23.

24. Muller-Eberhard, H. J., and R. D. Schreiber. 1980. Molecularbiology and chemistry of the alternative pathway of comple-ment. Adv. Immunol. 29:1-53.

25. Nuiens, J. H., J. J. Abbink, Y. T. Wachtfogel, R. W. Colman,A. J. M. Eerenberg, D. Dors, A. J. M. Kamp, R. J. M. Strackvan Schindel, L. G. Thus, and C. E. Hack. 1992. Plasmaal-antitrypsin and lactoferrin in sepsis: evidence for neutrophilsas mediators in fatal sepsis. J. Lab. Clin. Med. 119:159-168.

26. Nuoens, J. H., A. J. M. Eerenberg-Belmer, C. C. M. HuUbregts,W. 0. Schreuder, R. J. F. Felt-Bersma, J. J. Abbink, L. G. ThUs,and C. E. Hack. 1989. Proteolytic inactivation of plasma Cl-Inhibitor in sepsis. J. Clin. Invest. 84:443-450.

27. Okusawa, S., C. A. Dinarello, K. B. Yancey, S. Endres, T. J.Lawley, M. M. Frank, J. F. Burke, and J. A. Gelfand. 1987. CSainduction of human interleukin 1. Synergistic effect with endo-toxin or interferon-gamma. J. Immunol. 139:2635-2640.

28. Pinckard, R. N., M. S. Olson, P. C. Giclas, R. Terry, J. T.Boyer, and R. A. O'Rourke. 1975. Consumption of classicalcomplement components by heart subcellular membranes invitro and in patients after acute myocardial infarction. J. Clin.Invest. 56:740-750.

29. Sims, P. J., E. M. Faioni, T. Wiedmer, and S. J. Shattil. 1988.Complement proteins C5b-9 cause release of membrane vesiclesfrom the platelet surface that are enriched in the membrane

INFECT. IMMUN.

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

COMPLEMENT ACTIVATION IN E. COLI-CHALLENGED BABOONS 4301

receptor for coagulation factor Va and express prothrombinaseactivity. J. Biol. Chem. 263:18205-18212.

30. Starnes, H. F., Jr., M. K. Pearce, A. Tewari, J. H. Yim, J.-C.Zou, and J. S. Abrams. 1990. Anti-IL-6 monoclonal antibodiesprotect against lethal Escherichia coli infection and lethal tumornecrosis factor-a challenge in mice. J. Immunol. 145:4185-4191.

31. Stephens, K. E., A. Ishizaka, Z. Wu, J. W. Larrick, and T. A.Raffin. 1988. Granulocyte depletion prevents tumor necrosisfactor-mediated acute lung injury in guinea pigs. Am. Rev.Respir. Dis. 138:1300-1307.

32. Stevens, J. H., P. O'Hanley, J. M. Shapiro, F. G. Mihm, P. S.Satoh, J. A. Collins, and T. A. Raffin. 1986. Effects of anti-C5aantibodies on the adult respiratory distress syndrome in septicprimates. J. Clin. Invest. 77:1818-1826.

33. Taylor, F. B., Jr. Studies of the natural history and mechanismof the primate (baboon) response to E. coli. In D. C. Morrissonand J. L. Ryan (ed.), Bacterial endotoxic lipopolysaccharides,vol. 2, in press. CRC Press, Inc., Boca Raton, Fla.

34. Taylor, F. B., Jr., A. Chang, C. T. Esmon, A. D'Angelo, S.Vigano-D'Angelo, and K. E. Blick. 1987. Protein C prevents thecoagulopathic and lethal effects of Escherichia coli infusion inthe baboon. J. Clin. Invest. 79:918-925.

35. Taylor, F. B., Jr., A. Chang, G. Ferreil, T. Mather, R. Catlett,K. E. Blick, and C. T. Esmon. 1991. C4b-binding proteinexacerbates the host response to Escherichia coli. Blood 78:357-363.

36. Taylor, F. B., Jr., T. E. Emerson, Jr., R. Jordan, A. K. Chang,and K. E. Blick. 1988. Antithrombin-III prevents the lethal

effects of Escherichia coli infusion in baboons. Circ. Shock26:227-235.

37. Thus, L. G., C. E. Hack, R. J. M. Strack van Schijndel, J. H.Nuiens, G. J. Wolbink, A. J. M. Eerenberg-Belmer, H. van derVall, and J. Wagstaff. 1990. Activation of the complementsystem during immunotherapy with recombinant interleukin-2:relation to the development of side effects. J. Immunol. 144:2419-2424.

38. Vachino, G., J. A. Gelfand, M. B. Atkins, J. D. Tamerius, P.Demchak, and J. W. Mier. 1991. Complement activation incancer patients undergoing immunotherapy with interleukin-2(IL-2): binding of complement and C-reactive protein by IL-2-activated lymphocytes. Blood 78:2505-2513.

39. Viner, N. J., L. C. Bailey, P. F. Life, P. A. Bacon, and J. S. H.Gaston. 1991. Isolation of Yersinia-specific T cell clones fromthe synovial membrane and synovial fluid of a patient withreactive arthritis. Arthritis Rheum. 34:1151-1157.

40. Vogt, W. 1986. Anaphylatoxins: possible roles in disease. Com-plement 3:177-188.

41. Wiedmer, T., C. T. Esmon, and P. J. Sims. 1986. On themechanism by which complement proteins C5b-9 increase plate-let prothrombinase activity. J. Biol. Chem. 261:14587-14592.

42. Wolbink, G. J., M. Bollen, J. W. Baars, R. J. M. Ten Berge,A. J. G. Swaak, J. Paardekooper, and C. E. Hack. 1993.Application of a monoclonal antibody against a neoepitope onactivated C4 in an ELISA for the quantification of complementactivation via the classical pathway. J. Immunol. Methods163:67-76.

VOL. 61, 1993

on May 11, 2018 by guest

http://iai.asm.org/

Dow

nloaded from