THE ROLE OF COMPLEMENT ACTIVATION IN TUMOUR NECROSIS FACTOR-INDUCED LETHAL HEPATITIS

Article No. cyto.1998.0462, available online at http://www.idealibrary.com on

THE ROLE OF COMPLEMENT ACTIVATION IN TUMOUR NECROSIS FACTOR-INDUCED LETHAL HEPATITIS Claude Libert,1 Ben Wielockx,1 Bernardo Grijalba,1 Wim Van Molle,1 Elisabeth Kremmer,2 Harvey R. Colten,3 Walter Fiers,1 Peter Brouckaert1 Injection of tumour necrosis factor (TNF) in animals causes severe liver cell toxicity, especially when D-(+)-galactosamine (GalN) is co-administered. After challenge with TNF/GalN, serum complement activity (CH50 and APCH50) decreased dramatically, suggesting strong activation of both the classical and the alternative pathways. TNF or GalN alone had no such effect. A cleavage product of complement protein C3 [C3(b)] was deposited on the surface of hepatocytes of TNF/GalN-treated mice. Intravenous administration of cobra venom factor (CVF), which depletes complement, inhibited the development of hepatitis. However, CVF pretreatment also protected C3-deficient mice. Pretreatment of mice with a C1q-depleting antibody did not prevent TNF/GalN lethality, although the anti-C1q antibody had depleted plasma C1q. Factor B-deficient and C3-deficient mice, generated by gene targeting, proved to be as sensitive to TNF/GalN as control mice. Furthermore, induction of lethal shock by platelet-activating factor, an important mediator in TNF-induced hepatic failure, was not reduced in C3-deficient mice. These data indicate that complement, although activated, plays no major role in the generation of acute lethal hepatic failure in this model and that CVF-induced protection is independent of complement depletion.  1999 Academic Press

Tumour necrosis factor (TNF) is a pro-inflammatory cytokine, produced mainly by macrophages and exhibiting a broad range of biological activities.1 From a therapeutic point of view, the most interesting property is its antitumour effect, which is currently only applicable in isolated limb perfusion. Although TNF can destroy tumours in experimental animals and in regionally treated patients, the severe side effects, such as hypotension and liver toxicity after high doses in general circulation, prevent the treatment of cancer by systemic TNF administration.2,3 Understanding the From the 1Molecular Pathophysiology and Experimental Therapy Unit, Department of Molecular Biology, Flanders Interuniversity Institute for Biotechnology and University of Ghent, K.L. Ledeganckstraat 35, B-9000 Ghent, Belgium; 2GSFForschungszentrum fu¨r Umwelt und Gesundheit GmbH, Marchioninistraße 25, D-81377 Mu¨nchen, Germany; 3Department of Medicine, Washington University, 660 South Euclid Avenue, St. Louis, MO 63110, USA Correspondence to: Claude Libert Current address: Harvey R. Colten, Dean and V.P. for Medical Affairs, Northwestern University Medical School, Chicago, IL, USA Received 9 April 1998; accepted for publication 30 September 1998  1999 Academic Press 1043–4666/99/090617+09 $30.00/0 KEY WORDS: complement/hepatitis/knockout mice/shock/tumour necrosis factor CYTOKINE, Vol. 11, No. 8 (August), 1999: pp 617–625

molecular mechanism by which TNF induces liver toxicity is therefore of utmost importance in order to improve its therapeutic potential. Furthermore, TNF has been implicated as a mediator in various models of hepatitis and was also recognized as a central mediator in the endotoxin-induced systemic inflammatory response syndrome.4–6 Complement activation is an important step in propagating immune and inflammatory responses, involving the production of chemotactic polypeptides. Formation of several C3 split products can lead to activation of complement receptor-bearing leukocytes. The membrane attack complex not only lyses invading pathogens, but is also able to propagate the inflammatory response, e.g. by inducing the expression of tissue factor.7,8 Activation of the complement system was described in several model systems of inflammation and shock in which lipopolysaccharide (LPS) or TNF play a major role. In human patients it was demonstrated that liver cell necrosis is associated with complement activation.9 In order to increase the therapeutic value of TNF, we were interested to know to what extent complement activation constitutes an essential mechanism in the development of liver cell destruction by TNF. We used 617

618 / Libert et al.

CYTOKINE, Vol. 11, No. 8 (August, 1999: 617–625)

40

5000

120

A

4500 100

35

3000 2500 30

2000 1500 1000

25

CH50/APCH50 (%)

3500 AST (U/l)

Temperature (°C)

4000

80

60

40

500 1

2 3 4 6 5 Time after injection (h)

7

8

20

0

0

0

Figure 1. Metabolic response of mice after injection of a lethal dose of TNF/GalN.

1

2

3

4 5 6 7 Time after challenge (h)

8

9

Changes in body temperature ( ) (n=10) and serum AST ( ) (n=5) after injection of 0.5 ìg TNF+20 mg GalN in C57BL/6 mice.

B

80

CH50 (%)

a model in which TNF causes rapid irreversible hepatitis by co-injection with the hepatotoxic -(+)galactosamine (GalN). In this model, plateletactivating factor (PAF) is produced and plays an important role.10 Our results indicate that complement is activated and that C3 split products are deposited on hepatocytes, but these phenomena do not contribute to lethality.

100

60

40

RESULTS TNF/GalN-induced lethal hepatitis We first studied the LD100 of TNF/GalN and the changes of some metabolic parameters after injection with TNF/GalN. C57BL/6 mice were injected with increasing doses of TNF+20 mg GalN. The LD100 proved to be 0.1 ìg TNF+20 mg GalN per mouse (data not shown). 4 h after injection of 0.5 ìg TNF+20 mg GalN (5LD100), the body temperature started to drop, virtually to room temperature (Fig. 1). At the same time we observed an increase in liver aspartate aminotransferase (AST; Fig. 1), indicating fulminant necrosis. Tissue sections revealed that at the time death occurred (6–8 h after challenge) all liver cells were dead and massive amounts of erythrocytes were found in the liver. In this model, tissue damage was confined to the liver, since other organs remained unaffected.

CH50 and APCH50 measurements To measure complement activation in vivo after TNF/GalN administration, we studied changes in CH50 or APCH50. We made use of male BALB/c mice

20

0

Control

GalN

TNF

PBS TNF/GalN

Figure 2. Activation of complement by TNF/GalN in mice as measured by the decrease in CH50 and APCH50. (A) Drop in CH50 ( ) and APCH50 ( ) (%) as a function of time after injection of BALB/c mice (n=4) with 0.5 ìg TNF/GalN. CH50 and APCH50 of untreated mice were taken as 100%. (B) Effects on CH50 after injection of BALB/c mice (n=2) with 20 mg GalN, 0.5 ìg TNF, PBS or the combination of TNF/GalN. Mice were bled 7 h after injection.

as they proved to have higher complement activity in the serum than had female BALB/c or C57BL/6 mice (results not shown). Mice were given 0.5 ìg TNF/ GalN, about 5LD100. At the time points indicated in Figure 2A, mice were anaesthetized using avertin and bled by heart puncture. CH50 and APCH50 were determined as described in Materials and Methods. CH50 and APCH50 were expressed in percentage of untreated control mice (100%). From Figure 2A, we conclude that CH50 and APCH50 decrease gradually

Complement activation in hepatitis / 619

TABLE 1. Protective effect of CVF against TNF/GalNinduced transaminase release and lethality Pretreatment* PBS Mice C57BL/6 C30/0

TABLE 2. Response of C30/0 mice and wt mice to different doses of TNF/GalN Genotype

CVF

Lethality†

ALT‡

Lethality

ALT

8/8 9/9

1574915 32792189

1/8§ 1/7§

12279§ 352303§

*50 ìg CVF was injected i.v. 4 h prior to the challenge and again at the time of challenge, which consisted of 0.1 ìg TNF/20 mg GalN. †Number of dead mice vs total, scored 24 h after the challenge (no further deaths occurred). ‡U/l, measured 4 h after the challenge. §Statistically different from PBS control.

to low levels of about 20% 8 h after challenge. Activation of both the classical as well as the alternative pathways seems to occur. In a control experiment, mice were treated with 0.5 ìg TNF or 20 mg GalN or the combination of both; CH50 and APCH50 were measured 8 h after the injection. The results shown in Figure 2B demonstrate that only after injection of the combination of TNF/GalN did the CH50 levels decrease significantly. Comparable results were obtained in the case of APCH50 (not shown).

Protection by cobra venom factor (CVF)

C3 wt C30/0 C3 wt C30/0 C3 wt C30/0 C3 wt C30/0

Exp.

TNF dose*

Body temperature†

ALT‡

Lethality§

I I I I II II I I

0.03 0.03 0.1 0.1 0.1 0.1 0.3 0.3

37.60.8 37.30.7 33.73.4 36.61.7¶ ND ND ND ND

ND ND 300178 167170¶ ND ND ND ND

0/10 0/14 8/10 6/14¶ 5/10 5/9 7/9 7/9

*TNF (ìg/mouse) was administered i.p. in combination with 20 mg GalN. Injected volumes were adapted to the body weight. †C measured 8 h after the challenge. ‡U/l measured 6 h after the challenge. §Number of dead mice vs total scored 24 h after the challenge (no further deaths occurred). ¶Statistically different from wt mice.

the second experiment, no significant difference was observed. Furthermore, C30/0 mice were not protected against lethality induced by 0.3 ìg TNF/20 mg GalN. C30/0 mice had also released slightly less ALT 4 h after the challenge with 0.1 ìg TNF/GalN (experiment 1). Taken together, the results indicate that C30/0 mice are not significantly protected against TNF/GalN-induced lethal hepatitis.

We pretreated C57BL/6 mice with complementdepleting CVF as described previously.11 Mice were challenged with 0.1 ìg TNF/GalN and bled after 4 h, after which alanine aminotransferase (ALT) was determined in the serum. CVF protected against lethality and against release of transaminases (Table 1). However, when C30/0 mice were pretreated with CVF, they were also found to be protected against TNF/GalNinduced lethality and release of ALT. These data indicate that CVF protects against TNF/GalN-induced lethal hepatitis, but that the relevant activity of CVF is not by depleting C3.

We tested the effect of a rat anti-mouse C1q monoclonal antibody (mAb) on TNF/GalN-induced lethal hepatitis. The rat anti-mouse C1q antibody (Ab) RmC7H8 completely depleted C1q for at least 6 days in C57BL/6 mice after i.v. injection of 1 mg/mouse (Fig. 3), whereas the isotype control with anti-DNP had no influence on serum C1q (data not shown). As shown in Table 3, this dose of Ab, was not able to protect the animals against an LD50 or LD100 dose of TNF/GalN (Table 3).

Response of C30/0 mice to TNF/GalN

Contribution of alternative pathway activation

To study whether C3 was an essential constituent, we treated C30/0 mice (generated by gene targeting) with TNF/GalN. Offspring from couples of heterozygous mice were genotyped using PCR and a C3 ELISA. Homozygous deficient and wild-type (wt) mice were identified and further bred. The genotype of all mice used in the experiment was confirmed by ELISA. Mice were challenged with three different doses of TNF/GalN, namely 0.03 ìg TNF, 0.1 ìg TNF (2 experiments) or 0.3 ìg TNF (Table 2). The lowest dose of TNF/GalN was non-lethal in both control mice and C30/0 mice. A statistically better survival (P=0.034) was observed with C30/0 mice vs C3 wt mice at the higher dose of 0.1 ìg in one of the two experiments. In

To evaluate the role of the alternative pathway activation in the induction of lethal hepatitis by TNF/ GalN, we made use of factor B0/0 mice generated by gene targeting.12 Homozygous, deficient mice were used and compared with heterozygous mice which served as controls, since they are functionally similar to wt animals.12 Both types of mice were injected with 0.1 ìg TNF/GalN or with 0.5 ìg TNF/GalN. Serum was taken 6 h later, after which AST and ALT were determined (Fig. 4). For both the low dose and the high dose of TNF/GalN no significant differences in survival were observed. Factor B0/0 mice and control mice were equally sensitive to TNF/GalN-induced lethality (Fig. 4) as well as induction of serum ALT

Contribution of classical pathway activation

620 / Libert et al.

CYTOKINE, Vol. 11, No. 8 (August, 1999: 617–625)

100

100

A

90 80 % Survival

% C1q

80

60

70 60 50 40 30

40

20 10

20

0

0

20

40 60 Time after challenge (h)

80

100

0 0

10 15 20 Days after injection

25

30

100

C1q depletion in mice by mAb RmC7H8.

80

C1q depletion after injection of mAb RmC7H8 in different mouse strains. C1q levels of mice treated with an isotype control mAb were taken as 100%. Shown are mean values; n=4 for all groups. ( ), C57BL/6; ( ), BL/6CBA; (), CBA.

70 60 50 40

TABLE 3. Lack of protective effect of anti-C1q Ab against TNF/GalN-induced lethality

30

Pretreatment dose*

10

TNF dose†

Lethality after 24 h

B

90

% Survival

Figure 3.

5

Lethality after 48 h‡

20

0 Anti-C1q Ab PBS 1 mg Ab 1 mg control Ab PBS 1 mg Ab 1 mg control Ab

20 Time after challenge (h)

0.1 ìg 0.1 ìg 0.1 ìg 0.3 ìg 0.3 ìg 0.3 ìg

4/6 4/6 4/6 6/6 6/6 5/6

4/6 4/6 4/6 6/6 6/6 6/6

*C57BL/6 mice were pretreated i.p. 3 days before the challenge. Challenge was i.p. †Expressed in ìg/mouse in combination with 20 mg GalN. Volumes were adapted to the body weight. ‡No further deaths occurred.

(wt vs knockout; 10042 vs 10042 for 0.1 ìg TNF/GalN; 17801824 vs 14421442 for 0.5 ìg TNF/GalN) and AST (results not shown). The results suggest that the alternative pathway may be activated but is not contributing significantly to TNF/GalN-induced lethal hepatitis.

C3 deposition on hepatocytes 6 h after injection of 0.5 ìg TNF/GalN, tissue sections were prepared as described in Materials and Methods. C3 deposition was visualized by immunohistochemistry. We observed that C3 was present on the surface of the hepatocytes only after injection of TNF in combination with GalN (Fig. 5). No signal was observed after injection of TNF or GalN alone. The results indicate that a C3 fragment, probably C3b, was deposited on the hepatocytes.

Figure 4.

Lethal response of factor B0/0 mice to TNF/GalN.

Survival curves of factor B0/0 mice ( ) and control mice ( ) to a low dose of TNF/GalN (0.1 ìg; A) or a high dose of TNF/GalN (0.5 ìg; B). n=10 and n=14 for control and mutant mice, respectively (A); n=9 and n=13 for control and mutant mice, respectively (B).

PAF-induced lethal shock PAF was recognized as an important mediator in TNF/GalN-induced lethal hepatitis.10 Based on studies in naturally complement0/0 mice, PAF-induced lethal haemodynamic shock was reported to be depending on complement activation.13 Therefore the experiment recorded in Table 4 was performed. Increasing doses of PAF were administered i.p. in C30/0 and wt mice. From the results shown in Table 4, it is clear that there was no significant difference in response. Activation of complement does not seem to play an important role in PAF-induced lethal shock.

DISCUSSION TNF is a cytokine, produced mainly by activated macrophages upon stimulation with bacterial or viral products. It plays a central role in initiating the inflammatory response and is a beneficial factor in the

Complement activation in hepatitis / 621

Figure 5.

Deposition of C3b in the liver of TNF/GalN-treated mice.

Staining of liver sections from C57BL/6 mice injected with 0.5 ìg TNF+20 mg GalN for the presence of C3 (fragments). Livers were collected 6 h after injection. They were fixed and stained with (top) or without (bottom) anti-C3 polyclonal Ab.

defence against invading pathogens.1,14 Overproduction of TNF and release in the circulation, however, becomes dangerous and harmful as high concentrations of this cytokine can cause a systemic inflammatory response syndrome or shock.15 Furthermore, chronic (over)production of TNF has been found in various pathological conditions, such as systemic

inflammatory response syndrome, rheumatoid arthritis,16 inflammatory bowel disease,17 and adult respiratory distress syndrome.18 On the other hand, TNF also has antitumour activities. These are highly efficacious, especially when combined with interferon-ã.2 But the therapeutic use of TNF remains restricted to the protocol of isolated limb perfusion, because of the

622 / Libert et al.

CYTOKINE, Vol. 11, No. 8 (August, 1999: 617–625)

TABLE 4. Response of C30/0 mice and wt mice to PAFinduced lethal shock Genotype C3 wt C30/0 C3 wt C30/0 C3 wt C30/0 C3 wt C30/0 C3 wt C30/0

Dose PAF/mouse*

Lethality†

1 ìg

0/8 0/8 0/8 0/8 3/8 1/8 7/9 8/9 6/6 8/8

3 ìg 10 ìg 30 ìg 100 ìg

*PAF was injected i.p. in a volume of 0.5 ml. †Number of dead mice on a total scored 2 h after the challenge (no further deaths occurred).

aforementioned systemic toxicities. The major doselimiting toxicities of TNF as found in clinical and preclinical studies are hypotension and hepatotoxicity.19,20 Complement is a mediator activated by invading micro-organisms or their products. LPS has been shown to directly activate the complement cascade.21 After activation, a number of potent mediators is released, locally and in the circulation. C3a and C5a are anaphylatoxins with powerful biological activities, such as increase in vascular permeability, contraction of smooth muscle cells, attraction of inflammatory cells, induction of platelet aggregation and induction of broncho-constriction.22 In addition, both molecules have been shown to cause release of elastase from polymorphonuclear cells.22 C5a can also induce PAF release,23 enhance the synthesis of TNF by macrophages and activate polymorphonuclear cells.24 The membrane attack complex is not only involved in lysis of invading or infected cells, but also in propagating the inflammatory response. For instance, the membrane attack complex can induce expression of tissue factor and has other effects on coagulation.8,25 Using complement0/0 animals and inhibitors, complement activation has been observed and is believed to play a role in several model systems of shock and inflammation in which LPS and/or TNF are involved: lung injury,26 bowel injury and lethal shock,27 ischaemia–reperfusion-induced hepatic distress,28 pulmonary and hepatic increase in vascular permeability,23 LPS- or PAF-induced lethal shock,13 and induction of the Shwartzman reaction.11 We studied the mechanism by which TNF causes hepatic failure in a mouse model. Mice were treated i.p. with a combination of TNF and GalN. GalN is a sugar derivative which depletes uridine nucleotides and leads specifically to rapid arrest of translation in hepatocytes.29 The combination of TNF and GalN rapidly causes an irreversible and lethal necrosis of the liver

preceded by apoptosis of hepatocytes. After injection of TNF/GalN we observed that very high levels of transaminases were released. We have also found that PAF plays an important role as a mediator10 and that two acute phase proteins, namely á1-acid glycoprotein and á1-protease inhibitor protect against TNF/GalNinduced events.30,31 The protection by á1-protease inhibitor suggests an important role for one or more serine proteases in the pathology. In this study, we investigated to what extent complement activation is involved in TNF-induced lethal hepatitis, because of two reasons. First, hepatic failure is a major dose-limiting effect after TNF administration, and second, complement products were shown to play a role in experimental models of liver injury in rats and also in hepatic failure in human patients.9,32 A mediating role of complement is an obvious hypothesis and, if confirmed, would contribute to our knowledge of the molecular events leading to TNF-induced lethal shock and would allow a broadening of the therapeutic value of TNF. The results here reported indicate that complement indeed becomes activated after injection of TNF/GalN. First, we observed a marked reduction in CH50 and APCH50. These data indicate that complement components had been depleted by activation of both complement pathways. Alternatively, complement components may be cleaved by an overwhelming increase in serum protease activity, as suggested by á1-antitrypsin protection against TNF/GalN-induced lethal hepatitis.31 The GalN control suggests that the effects of TNF/GalN were not the result of a sudden arrest of production of complement factors by the hepatocytes. Besides, the half-lives in circulation of the complement components are very long, of the order of a few days.33 Second, when we made sections of the liver after injection of TNF/GalN, we observed a strong signal using an Ab to mouse C3. Very probably this indicates deposition of C3b on the surface of the hepatocytes. It has been described that C3 deposition and subsequent necrosis in human liver occurred only after one of the protective factors, viz. membrane co-factor protein, had disappeared at those necrotic foci.9 We have indications that the same mechanism happens in the mouse model, since crry/p65, the major restriction factor in mice, is downmodulated by GalN (manuscript in preparation). Third, treatment with CVF, which depletes complement, protected against TNF/GalN induced lethality and transaminase release. Based on the protection induced by CVF, we expected C30/0 mice to be resistant to TNF/GalN. Using such mice, generated by gene targeting and completely devoid of any complement activity, however, we found no significant protection. C30/0 and wt mice were derived from the same heterozygous breeding couples so that they differed in the function of the C3 gene

Complement activation in hepatitis / 623

only. CVF also protected C30/0 mice against TNF/ GalN, so that the data suggest that protection by CVF is not complement dependent. CVF protects against TNF/GalN by a mechanism still unidentified. Further studies indicated that complement activation is not causally related to the induction of lethal hepatitis by TNF. Neither a mAb entirely depleting C1q nor a very high dose of human C1 inhibitor (results not shown) protected against lethal or sublethal amounts of TNF/GalN. Furthermore, using factor B0/0 mice, generated by gene targeting, we were unable to attribute a mediating role to activation of the alternative pathway. These data are in line with our findings that even a high dose of recombinant soluble CR1, which in the mouse inhibits the alternative pathway, was unable to protect against TNF/GalN (results not shown). PAF, a central mediator of TNF-induced hepatitis, causes rapid and lethal hemodynamic shock when injected in animals. We found that C30/0 mice respond like wt litter mates to sublethal or lethal doses of PAF. These data are in line with expectations based on the aforementioned results with TNF/GalN in C30/0 mice, but are in contrast to previous reports that describe decreased sensitivity of partially C3-deficient mice to PAF.13 We observed an analogous partial protection by partially C3-deficient strains of mice to TNF/GalN (DBA/2 vs CBA/J and C57BL/10SnJ vs C57BL/6 or C57Bl/10J; results not shown). These differences in response are likely to be the result of other genetic polymorphisms. The data here reported indicate that complement activation clearly occurs but that it plays no significant role in the TNF/GalN pathogenesis. We conclude that inhibition of complement activation will not contribute to an increase in the therapeutic value of TNF.

MATERIALS AND METHODS Animals Female C57BL/6, CBA and (C57BL/6CBA)F1, as well as male BALB/c mice were purchased from Charles River (Broekman Instituut, Someren, The Netherlands). Factor B0/0 mice12 and C30/0 mice (H.R. Colten, manuscript in preparation) were generated by gene targeting as described. Mice were kept in 12-h light/dark cycles in a temperaturecontrolled, air-conditioned room and received food and water ad libitum. They were used at the age of 8–10 weeks. Rectal body temperatures were measured with an electronic thermometer (model 2001; Comark Electronics, Littlehampton, UK).

Reagents GalN, extrAvidin, fast red substrate, BSA, PAF, CVF from Naja naja kaouthia, a goat anti-rat Ab and goat serum were from Sigma Chemical Co. (St Louis, MO). Murine

recombinant TNF was produced in our laboratory; it had a specific activity of 1.3108 IU/mg and contained <1.8 ng endotoxin/mg protein. Goat anti-mouse C3 was obtained from ICN Pharmaceuticals (Costa Mesa, CA). A mouse anti-rabbit erythrocyte polyclonal Ab was generously provided by Dr H. van Dijk (University of Utrecht, The Netherlands). A rat anti-mouse C3 mAb (RmC11H9) and a rat anti-mouse C1q mAb (RmC7H8) were generated and produced in our laboratories (E.K.).34 Endotoxin levels were assessed by a chromogenic Limulus amoebocyte lysate assay (Coatest; Chromogenix, Stockholm, Sweden).

C1q depletion by RmC7H8 C57BL/6, CBA and (C57BL/6CBA)F1 mice were i.v. given protein G-purified RmC7H8 of rat IgG1 subclass (1 mg/250 ìl PBS). Tail vein blood was taken in the first week every second day, then twice a week up to normalization of C1q. Untreated mice or mice treated with an irrelevant anti-DNP rat IgG1 mAb (a gift of Dr H. Bazin, Universite´ Catholique de Louvain, Brussels, Belgium) served as controls. Sera were tested for C1q using an mAb against C1q detecting a second epitope on C1q (RmC4B9; rat IgM). To this end, ELISA plates were coated with 5 ìg/ml goat F(ab)2 anti-rat IgM (ì chain-specific). After blocking the plates with non-fat milk, supernatant of RmC4B9 was added. Next, serial dilutions of sera were incubated in excess RmC7H8. Bound RmC7H8 was detected using a biotinylated mouse mAb against rat IgG1 (TIB 170; ATCC, Rockville, MD) and horseradish peroxidase avidin D (Vector Laboratories, Burlingame, CA). The percentage of C1q was calculated on the basis of C1q titers of mice treated with irrelevant mAb.

Injections and blood collections All reagents were diluted in pyrogen-free PBS immediately before injection. The i.p. injections had a volume of 0.5 ml; the i.v. injections were in a volume of 0.25 ml. In most experiments, TNF doses were adapted to the body weight of the animals. Blood was collected by retro-orbital bleeding under ether anaesthesia or by heart puncture under tribromoethanol anaesthesia. Serum was prepared by incubation of the blood samples for 30 min at 37C, separation of the fibrin clot, and centrifugation for 15 min at 16 000g. Serum samples were stored at
80C before use. PAF was kept in CHCl3 at
20C, dried and dissolved in PBS containing 0.25% BSA prior to injection.

ALT and AST quantification Serum ALT and AST were determined using a Hitachi analyser and an ALT/AST kit from Sigma Chemical Co.

Measurements of CH50 and APCH50 CH50 and APCH50 were measured as described previously.35 Briefly, serum samples were titrated 1/2 in veronal saline buffer containing CaCl2 (CH50) or EGTA (APCH50) in 96-well plates. Ab-sensitized rabbit erythrocytes were added and the plates were incubated for 1 h at 30C (CH50) or at 37C (APCH50). After incubation, the plates were centrifuged and supernatants were transferred to a new plate and read at 405 nm. To exclude interference by serum

624 / Libert et al.

hemoglobin, all samples were also tested after heat inactivation (30 min at 56C). CH50 and APCH50 values were calculated using H20 and veronal saline buffer + + controls as described.

Immunohistochemistry on liver sections Mice were killed by cervical dislocation and the abdomen was opened. The liver was perfused using 10 ml PBS and 10 ml of a 2% paraformaldehyde solution. After washing, the livers were kept in 0.5 M sucrose, dried and stored at
80C. Sections of 8 ìm were cut at
20C. The sections were treated with avidin (30 min at room temperature) and with biotin (10 min at room temperature). After washing, the sections were blocked with 10% goat serum (30 min at room temperature) and the first Ab, a rat anti-mouse C3, was applied in 10% goat serum. After incubation, biotin-labelled goat anti-rat was added. C3 was visualized using extrAvidin alkaline phosphatase and fast red as a substrate.

Statistics The significance of differences in body temperature, AST and ALT was calculated using a two-tailed t-test or an ANOVA t-test. Significant differences in final survival were calculated using a ÷2 test, and significant differences in survival time (Kaplan–Meier plots) were calculated using the Logrank test.

Acknowledgements The authors thank A. Raeymaekers for purification of TNF and L. Van Geert for technical assistance. Drs P. Gronski, G. Dickneite, H. Marsh and H. van Dijk are acknowledged for providing reagents. We thank Dr. G. Leroux-Roels for providing AST and ALT determination facilities. B.W. is a fellow with the Vlaams Instituut voor de Bevordering van het Wetenschappelijk-technologisch Onderzoek in de Industrie. WVM is a research assistant and PB a research associate with the Fonds voor Wetenschappelijk Onderzoek-Vlaanderen. Research was supported by the Fonds voor Wetenschappelijk OnderzoekVlaanderen (grant G023698N), the Algemene Spaaren Lijfrentekas, the Interuniversitaire Attractiepolen, and the US Public Health Services (grants A124836, A124739 and HD17461).

REFERENCES 1. Fiers W (1995) Biologic therapy with TNF: Preclinical studies. In DeVita VT Jr, Hellman S, Rosenberg SA (eds) Biologic Therapy of Cancer, 2nd edn., J.B. Lippincott, Philadelphia, pp. 295– 327. 2. Brouckaert PGG, Leroux-Roels GG, Guisez Y, Tavernier J, Fiers W (1986) In vivo anti-tumour activity of recombinant human and murine TNF, alone and in combination with murine IFNgamma, on a syngeneic murine melanoma. Int J Cancer 38:763–769. 3. Lejeune F, Lie´nard D, Eggermont A, Schraffordt Koops H, Kroon B, Ge´rain J, Rosenkaimer F, Schmitz P (1994) Clinical

CYTOKINE, Vol. 11, No. 8 (August, 1999: 617–625) experience with high-dose tumor necrosis factor alpha in regional therapy of advanced melanoma. Circ Shock 43:191–197. 4. Leist M, Gantner F, Jilg S, Wendel A (1995) Activation of the 55 kDa TNF receptor is necessary and sufficient for TNFinduced liver failure, hepatocyte apoptosis, and nitrite release. J Immunol 154:1307–1316. 5. Leist M, Gantner F, Bohlinger I, Tiegs G, Germann PG, Wendel A (1995) Tumor necrosis factor-induced hepatocyte apoptosis precedes liver failure in experimental murine shock models. Am J Pathol 146:1220–1234. 6. Brouckaert P, Fiers W (1996) Tumor necrosis factor and the systemic inflammatory response syndrome. Curr Topics Microbiol Immunol 216:167–187. 7. Esser AF (1991) Big MAC attack: Complement proteins cause leaky patches. Immunol Today 12:316–318. 8. Saadi S, Holzknecht RA, Patte CP, Stern DM, Platt JL (1995) Complement-mediated regulation of tissue factor activity in endothelium. J Exp Med 182:1807–1814. 9. Pham BN, Mosnier JF, Durand F, Scoazec JY, Chazouille`res O, Degos F, Belghiti J, Degott C, Benhamou JP, Erlinger S (1995) Immunostaining for membrane attack complex of complement is related to cell necrosis in fulminant and acute hepatitis. Gastroenterology 108:495–504. 10. Libert C, Van Molle W, Brouckaert P, Fiers W (1996) Platelet-activating factor is a mediator in tumor necrosis factor/ galactosamine-induced lethality. J Inflam 46:139–143. 11. Rothstein JL, Lint TF, Schreiber H (1988) Tumor necrosis factor/cachectin. Induction of hemorrhagic necrosis in normal tissue requires the fifth component of complement (C5). J Exp Med 168:2007–2021. 12. Matsumoto M, Fukuda W, Circolo A, Goellner J, StraussSchoenberger J, Wang X, Fujita S, Hidvegi T, Chaplin DD, Colten HR (1997) Abrogation of the alternative complement pathway by targeted deletion of murine factor B. Proc Natl Acad Sci USA 94:8720–8725. 13. Sun X, Hsueh W (1991) Platelet-activating factor produces shock, in vivo complement activation, and tissue injury in mice. J Immunol 147:509–514. 14. Vassalli P (1992) The pathophysiology of tumor necrosis factors. Annu Rev Immunol 10:411–452. 15. Kettelhut IC, Fiers W, Goldberg AL (1987) The toxic effects of tumor necrosis factor in vivo and their prevention by cyclooxygenase inhibitors. Proc Natl Acad Sci USA 84:4273–4277. 16. Keffer J, Probert L, Cazlaris H, Georgopoulos S, Kaslaris E, Kioussis D, Kollias G (1991) Transgenic mice expressing human tumour necrosis factor: A predictive genetic model of arthritis. EMBO J 10:4025–4031. 17. Kojouharoff G, Hans W, Obermeier F, Ma¨nnel DN, Andus T, Scholmerich J, Gross V, Falk W (1997) Neutralization of tumour necrosis factor (TNF) but not of IL-1 reduces inflammation in chronic dextran sulphate sodium-induced colitis in mice. Clin Expl Immunol 107:353–358. 18. Roten R, Markert M, Feihl F, Schaller MD, Tagan MC, Perret C (1991) Plasma levels of tumor necrosis factor in the adult respiratory distress syndrome. Am Rev Respir Dis 143:590– 592. 19. Brouckaert P, Libert C, Cauwels A, Everaerdt B, Van Molle W, Ameloot P, Gansemans Y, Grijalba B, Takahashi N, Truong M-J, Van Leuven P, Fiers W (1996) Inhibition of and sensitization to the lethal effects of tumor necrosis factor. J Inflam 47:18–26. 20. Zhang DF, Ren H, Jia XP, Zhou YS (1993) Serum tumor necrosis factor (TNF) in the pathogenesis of clinical hepatic failure of HCV and/or HBV. Chin Med J Engl 106:335–338. 21. de Boer JP, Creasey AA, Chang A, Roem D, Eerenberg AJM, Hack CE, Taylor FB Jr (1993) Activation of the complement system in baboons challenged with live Escherichia coli: Correlation with mortality and evidence for a biphasic activation pattern. Infect Immun 61:4293–4301. 22. Frank MM, Fries LF (1991) The role of complement in inflammation and phagocytosis. Immunol Today 12:322–326. 23. Lianos EA, Zanglis A (1992) Effects of complement activation on platelet-activating factor and eicosanoid synthesis in rat mesangial cells. J Lab Clin Med 120:459–464.

Complement activation in hepatitis / 625 24. Barton PA, Warren JS (1993) Complement component C5 modulates the systemic tumor necrosis factor response in murine endotoxic shock. Infect Immun 61:1474–1481. 25. Hamilton KK, Zhao J, Sims PJ (1993) Interaction between apolipoproteins A-I and A-II and the membrane attack complex of complement. Affinity of the apoproteins for polymeric C9. J Biol Chem 268:3632–3638. 26. Rabinovici R, Yeh CG, Hillegass LM, Griswold DE, DiMartino MJ, Vernick J, Fong K-LL, Feuerstein G (1992) Role of complement in endotoxin/platelet-activating factor-induced lung injury. J Immunol 149:1744–1750. 27. Hsueh W, Gonzalez-Crussi F (1988) Ischemic bowel necrosis induced by platelet-activating factor: An experimental model. Methods Achiev Exp Pathol 13:208–239. 28. Jaeschke H, Farhood A, Bautista AP, Spolarics Z, Spitzer JJ (1993) Complement activates Kupffer cells and neutrophils during reperfusion after hepatic ischemia. Am J Physiol 264:G801–G809. 29. Decker K, Keppler D (1974) Galactosamine hepatitis: Key role of the nucleotide deficiency period in the pathogenesis of cell injury and cell death. Rev Physiol Biochem Pharmacol 71:77–106.

30. Libert C, Brouckaert P, Fiers W (1994) Protection by á1-acid glycoprotein against tumor necrosis factor-induced lethality. J Exp Med 180:1571–1575. 31. Libert C, Van Molle W, Brouckaert P, Fiers W (1996) á1-Antitrypsin inhibits the lethal response to TNF in mice. J Immunol 157:5126–5129. 32. Matsuo R, Ukida M, Nishikawa Y, Omori N, Tsuji T (1992) The role of Kupffer cells in complement activation in D-galactosamine/lipopolysaccharide-induced hepatic injury of rats. Acta Med Okayama 46:345–354. 33. Barnum SR, Fey G, Tack BF (1989) Biosynthesis and genetics of C3. Curr Top Microbiol Immunol 153:23–43. 34. Wasiliu M, Kremmer E, Thierfelder S, Felber E, HoffmannFezer G, Kummer U (1989) Monoclonal antibodies to complement components without the need of their prior purification. I. Antibodies to mouse C1q. Hybridoma 8:615–622. 35. Klerx JPAM, Beukelman CJ, Van Dijk H, Willers JMN (1983) Microassay for colorimetric estimation of complement activity in guinea pig, human and mouse serum. J Immunol Methods 63:215–220.