Bimodal role of endogenous interleukin-6 in concanavalin A-induced ...

Bimodal role of endogenous interleukin-6 in concanavalin A-induced hepatitis in mice Yoh-ichi Tagawa,* Patrick Matthys,* Hubertine Heremans,* Chris Dillen,* Zahur Zaman,† Yoichiro Iwakura,‡ and Alfons Billiau* *Laboratory of Immunobiology, Rega Institute, University of Leuven; †Division of Experimental Laboratory Medicine, University Hospital Gasthuisberg, Leuven, Belgium; and ‡Laboratory Animal Research Center, Institute of Medical Science, University of Tokyo, Japan

Abstract: Acute concanavalin A (Con A)-induced hepatitis in mice is an animal model for hepatic injury induced by activated T cells. The evolution of hepatic involvement can be followed from hour to hour by measuring serum transaminase levels. We investigated the possible role of endogenous interleukin-6 (IL-6) in this model. We found serum IL-6 levels and splenic IL-6 mRNA during Con A-induced hepatitis to be significantly lower in interferon-␥ (IFN-␥)-deficient mice, which are resistant against the Con A-induced syndrome, than in wild-type ones, suggesting that systemic IL-6 production favors development of hepatic injury. However, IL-6-deficient mice proved to be more susceptible to the disease than wild-type mice, indicating that endogenous IL-6 plays a predominantly hepatoprotective role. Experiments in which wild-type mice were treated with anti-IL-6 antibodies, before or after Con A challenge, allowed us to reconcile these contrasting observations. The antibody injections resulted in a biphasic alteration of serum IL-6 levels, initial neutralization being followed by rebound increased levels due to accumulation of IL-6 in the form of antigen-antibody complexes. The effect of antibody on disease severity differed depending on the time of injection. Antibody injection at 2.5 h post Con A resulted in delayed disease manifestation, whereas treatment initiated before Con A resulted in accelerated disease. We conclude that endogenous IL-6 plays a bimodal role. IL-6 present before Con A challenge as well as that induced in the very early phase after Con A injection triggers hepatoprotective pathways. Continuation of IL-6 production beyond this early phase, by some other pathway, seems to be harmful to hepatocytes. J. Leukoc. Biol. 67: 90–96; 2000. Key Words: interferon-␥ hepatic injury

(TNF), and interleukin-6 (IL-6)] into the circulation accompanied by liver injury consisting of hepatocyte apoptosis and infiltration of the parenchyma with mononuclear cells [1, 2]. This Con A-induced hepatitis syndrome is considered useful as a model for the pathogenesis of immune-mediated liver damage occurring in humans infected with hepatitis viruses. The role of individual cytokines in causing and controlling this disease is under investigation [3–8]. IFN-␥-deficient mice have been reported to be resistant to the Con A-induced syndrome [7], and treatment of sensitive mice with neutralizing anti-IFN-␥ has been found to provide protection [4, 6], thus indicating that IFN-␥, produced as a result of the Con A challenge, acts as a strong contributor to the pathogenesis. The role of IFN-␥ seems to consist mainly in activating the Fas/FasL system in the liver, resulting in hepatocyte apoptosis [7, 9–11]. The role of other cytokines is less well established. IL-6 is of particular interest because it stimulates hepatocytes to produce acute-phase reactants, some of which have been found to protect hepatocytes against damage [12, 13]. Available evidence concerning the effects of IL-6 in the Con A-induced syndrome is equivocal: Mizuhara et al. [2, 6] found that recombinant IL-6 injection before Con A suppressed disease, whereas administration subsequent to Con A resulted in more severe liver injury. Treatment with anti-IL-6 antibody failed to affect the disease. To obtain more direct evidence for a role of endogenous IL-6 in Con A-induced hepatitis, we compared the susceptibility of IL-6-deficient mice with that of wild-type mice and we tested the effect of various schedules of treatment with anti-IL-6 antibodies. We considered these two approaches to complement each other because the IL-6 gene knockout approach leaves no doubt as to whether all IL-6 is completely ablated, whereas the anti-IL-6 antibody approach allows one to study the effect of IL-6 neutralization at particular times during the disease process. Our results indicate that IL-6 presence before and/or in the early phase of the disease is critical for evoking a strong hepatoprotective effect, and that continued high levels of IL-6 are harmful for the liver.

INTRODUCTION Injection of mice with the T-cell mitogen, concanavalin A (Con A), results in a massive release of several cytokines [e.g., interferon-␥ (IFN-␥), interleukin-1 (IL-1), tumor necrosis factor 90

Journal of Leukocyte Biology Volume 67, January 2000

Correspondence: Dr. Alfons Billiau, Laboratory of Immunobiology, Rega Institute, University of Leuven, Minderbroedersstraat 10, B-3000 Leuven, Belgium. E-mail: [email protected] Received July 6, 1999; revised September 27, 1999; accepted September 30, 1999.

http://www.jleukbio.org

random-primed 32P-labeled cDNA probe (1 ⫻ 106 dpm/mL) for 18 h at 42°C. Murine IL-6 cDNA was kindly provided by Dr. Takashi Yokota (Institute of Medical Science, Tokyo, Japan) [19]; the Bam HI (1.2 kb) fragment was used.

MATERIALS AND METHODS Mice The IFN-␥-deficient BALB/cA mouse substrain was obtained by six backcrosses of the original IFN-␥-deficient 129 ⫻ BALB/c strain [7] to the parental BALB/cA strain. Wild-type BALB/c mice were purchased from Bantin and Kingman (North Humberside, UK). IL-6-deficient C57BL/6 mice were backcrosses to C57BL/6 from the original 129 ⫻ C57BL/6-deficient mice generated by Kopf et al. [14]. These mice were obtained in second generation through the courtesy of Dr. Kopf (Basel Institute of Immunology, Basel, Switzerland) or were purchased as eighth generation from Jackson Laboratory (Bar Harbor, ME). All experiments were done on male mice of 9–14 weeks of age. Animal experiments were conducted according to institutional ethical guidelines. Intravenous injections were given in the tail vein. Blood samples were taken from the orbital sinus under ether anesthesia.

Reagents A stock solution of Con A (Sigma Chemical, St. Louis, MO) was kept at 2 mg/mL and was appropriately diluted for injections. The monoclonal anti-IL-6 antibodies, 6B4 [15] and 20F3 [16], were prepared from the corresponding hybridomas, grown in Pristane-primed NMRI nu/nu mice. The antibodies were purified by affinity chromatography on a mouse anti-rat ␬-chain antibodySepharose column (antibody MARK-1, provided by Dr. H. Bazin, Universite Catholique de Louvain, Belgium).

Assays Blood samples were kept at 4°C for 8 h before centrifugation (3,000 rpm, 10 min) to harvest serum. Although alkaline phosphatase (ALP) levels were determined according to the optimized method recommended by the German Society of Clinical Chemistry, alanine transaminase (ALT) and aspartate transaminase (AST) levels were measured by IFCC (International Federation of Clinical Chemists) procedures, using a Boehringer Mannheim/Hitachi 917 apparatus (Boehringer Mannheim, Mannheim, Germany). IL-6 levels were determined by a bioassay based on the growth-promoting effect for 7TD1 hybridoma cells [17]. One unit per milliliter of IL-6 was defined as the concentration that resulted in half-maximal growth of cells. When transaminase or IL-6 data of independent experiments were pooled (see Figs. 2 and 3), the values in each experiment were standardized using a correction factor, c ⫽ (mean value of untreated mice in all pooled experiments)/(mean value of untreated mice in the experiment concerned). Total RNA was isolated by the acid guanidinium isothiocyanate-phenol-chloroform/isoamylalcohol method [18]. RNA samples (5–8 µg) were electrophoresed on 1.2% denaturing agarose-formaldehyde gels. The separated RNA bands were transferred to nylon filters (GeneScreenPlus, DuPont-NEN, Boston, MA) by capillary blotting, followed by ultraviolet cross-linking. Hybridization was carried out with a

TABLE 1.

2

0

20

30

20

20

Circulating levels of IL-6 bioactivity after Con A injection in hepatitis-resistant IFN-␥-deficient and in hepatitis-sensitive wild-type mice As a first approach to see whether IL-6 production is related to severity of the Con A-induced hepatitis syndrome, we compared Con A-induced serum transaminase and IL-6 levels in wild-type and IFN-␥-deficient mice, the latter being known to be more resistant to the syndrome. Groups of IFN-␥-deficient or wild-type BALB/c mice were injected with saline, or with 20 or 30 µg/g body weight Con A. Transaminase and IL-6 levels were determined in serum samples taken 20 or 24 h after injection. As can be seen in Table 1, the transaminase levels were significantly lower in the mutant than in the wild-type mice. In parallel, IL-6 levels were also significantly lower in the IFN-␥-deficient mice. The difference in IL-6 response between IFN-␥-deficient and IFN-␥-competent mice was minimal or nonexistent in the early phase after Con A injection (time 1.5 h), but significant and sustained in the later period (time points 3, 6, and 24 h). These results raise the question of whether the disease-promoting effect of IFN-␥ in the Con A model is mediated by augmented production of IL-6, or alternatively, whether high-level IL-6 production is just a symptom of the systemic inflammatory response.

IL-6-deficient mice are more susceptible to Con A-induced hepatitis As a straightforward approach to investigate the possible role of IL-6 in Con A-induced hepatitis, we wanted to compare IL-6 gene knockout mice with wild-type mice for their sensitivity to Con A-induced hepatitis. Available IL-6 knockout mice were of C57BL/6 back-ground. In such mice, as in BALB/c mice, ablation of IFN-␥ results in a decreased hepatotoxic response to

Serum Transaminase and IL-6 Levels in IFN-␥-Deficient and Wild-Type Mice After Injection of Con A

Blood sample time Experiment Con A dose (h post number (µg/g, i.v.) Con A)

1

RESULTS

1.5 3 6 24

Serum level of Mouse genotype (BALB/c)

IL-6 (log10 U/mL)

AST (U/L)

ALT (U/L)

ALP (U/L)

IFN-␥ KO ⬍0.5 (N ⫽ 4) 259 ⫾ 123 (N ⫽ 4) 95 ⫾ 49 (N ⫽ 4) 450 ⫾ 76 (N ⫽ 4) wild-type ⬍0.5 (N ⫽ 4) 251 ⫾ 120 (N ⫽ 4) 79 ⫾ 55 (N ⫽ 4) 603 ⫾ 197 (N ⫽ 4) IFN-␥ KO 1.82 ⫾ 0.03 (N ⫽ 4)** 1,386 ⫾ 317 (N ⫽ 4)** 2,092 ⫾ 493 (N ⫽ 4)** 940 ⫾ 280 (N ⫽ 4)* wild-type 3.25 ⫾ 0.26 (N ⫽ 3) 8,929 ⫾ 1,551 (N ⫽ 3) 14,409 ⫾ 1498 (N ⫽ 3) 1,447 ⫾ 47 (N ⫽ 3) IFN-␥ KO wild-type IFN-␥ KO wild-type IFN-␥ KO wild-type IFN-␥ KO wild-type

2.45 ⫾ 0.12 (N ⫽ 4) 174 ⫾ 24 (N ⫽ 5) 2.18 ⫾ 0.05 (N ⫽ 4) 112 ⫾ 6 (N ⫽ 5) 3.03 ⫾ 0.05 (N ⫽ 4) 204 ⫾ 37 (N ⫽ 5) 3.40 ⫾ 0.21 (N ⫽ 4) 156 ⫾ 27 (N ⫽ 5) 2.38 ⫾ 0.05 (N ⫽ 4) 536 ⫾ 97 (N ⫽ 5) 3.03 ⫾ 0.08 (N ⫽ 4) 382 ⫾ 51 (N ⫽ 5) 1.95 ⫾ 0.12 (N ⫽ 4)** 180 ⫾ 25 (N ⫽ 11)** 2.63 ⫾ 0.16 (N ⫽ 4) 1,625 ⫾ 264 (N ⫽ 12)

84 ⫾ 16 (N ⫽ 5) 77 ⫾ 21 (N ⫽ 5) 286 ⫾ 116 (N ⫽ 5) 90 ⫾ 31 (N ⫽ 5) 704 ⫾ 235 (N ⫽ 5) 416 ⫾ 147 (N ⫽ 5) 373 ⫾ 73 (N ⫽ 11)** 2,743 ⫾ 426 (N ⫽ 12)

n.d. n.d. n.d. n.d. n.d. n.d. 213 ⫾ 16 (N ⫽ 11)* 478 ⫾ 68 (N ⫽ 12)

Statistical analysis (Student’s t test): * P ⬍ 0.005, ** P ⬍ 0.0005 for comparison with wild-type.

Tagawa et al. Interleukin-6 in concanavalin A-induced hepatitis

91

Con A [7]. However, in addition, we verified that as in BALB/c mice, ablation of IFN-␥ also resulted in lesser production of IL-6. This was demonstrated by comparing Con A-induced levels of IL-6 mRNA in spleens and livers. As can be seen in Figure 1, the organs of homozygous IFN-␥-deficient mice contained markedly less mRNA for IL-6 than those of heterozygous mice. Hence, since Con A-injected BALB/c and C57BL/6 mice did not differ in terms of the effects of endogenous IFN-␥ on Con A-induced IL-6 production and hepatotoxicity, we proceeded to compare groups of IL-6-deficient or wild-type C57BL/6 mice for Con A-induced hepatitis. Blood was taken from individual mice 20 h after intravenous injection with 20 µg/g body weight Con A. As shown in Table 2, transaminase levels were significantly higher in Con A-treated IL-6-deficient mice than in the wild-type mice, suggesting that the role of IL-6 in the Con A-induced hepatitis syndrome is to protect the liver against damage.

Influence of anti-IL-6 antibody on development of Con A-induced hepatitis As a second approach to investigate the role of IL-6 in the Con A model, we used a neutralizing anti-IL-6 monoclonal antibody (20F3). From previous studies [17, 20–22] it is known that in vivo administration of anti-IL-6 antibodies results in accumulation of antigen-antibody complexes from which biologically active IL-6 can be released, so that effective neutralization is only transient. Therefore, we designed antibody injection schedules aiming at neutralizing IL-6 in selected short time intervals after Con A injection. The antibody was injected in BALB/c mice at 0, 2.5, or 5 h after Con A injection. Blood samples for transaminase level determination were taken at 5, 7, or 20 h after Con A, as indicated in Table 3. Final (20 h) levels of transaminases remained unaffected by the anti-IL-6 antibody treatments. However, 5-h transaminase levels were significantly reduced when antibody injections were given 2.5 h post Con A but not, or only marginally so, when they were administered simultaneously with Con A. Reduction of the transaminase levels in mice receiving the effective antibody treatment was associated with reduced IL-6 levels (Table 3). We did a time-kinetic analysis of IL-6 biological activity in mice receiving the effective anti-IL-6 antibody injection at 2.5 h post Con A. As shown in Figure 2A, the biological activity first decreased in the antibody-treated mice, but within 3 h after injection of the antibody, it increased again to reach a level that was 100-fold higher than in control IgG-treated mice. Concomi-

Fig. 1. IL-6 mRNA levels in the spleen and liver of Con A-treated IFN-␥-deficient mice (C57BL/6J). Total RNA was prepared from the spleen or liver removed 12 h after Con A injection (30 µg/g body weight). IL-6 mRNA was detected by Northern blot hybridization analysis. The same filter was probed with ␤-actin to normalize amounts of RNA that were loaded.

92


TABLE 2. Serum Transaminase Levels in IL-6-Deficient and Wild-Type Mice Given a Single Injection of Con Aa Transaminase level (U/L) Experiment number

Genotype

N

AST

ALT

1

IL-6 KO wild-type

5 3

2,564 ⫾ 620* 360 ⫾ 12

4,130 ⫾ 1,000* 420 ⫾ 45

2

IL-6 KO wild-type

7 4

4,334 ⫾ 553** 1,058 ⫾ 698

5,514 ⫾ 553** 1,448 ⫾ 1,050

a IL-6-deficient mice of C57BL/6 background (F2 and F8 generation in experiment 1 and 2, respectively). Con A, 20 µg/g body weight, i.v.; Student’s t test for difference with wild-type, * P ⬍ 0.025, ** P ⬍ 0.001.

tantly, elevation of ALT levels was delayed in the antibodytreated mice (Fig. 2B). This indicated that the potential of anti-IL-6 antibody treatment to neutralize endogenous IL-6 and hence to affect the disease process is of very short duration and that, therefore, the time point of their administration is critical. To obtain further support for this transient protective effect of anti-IL-6 antibody we tested whether an additional injection of the antibody could suppress transaminase levels for a longer time. Antibody injections were therefore given at 2.5 and 5.5 h after Con A injection. Blood samples were taken at 8 and 20 h after Con A challenge. As can be seen in Table 4, two successive injections of 20F3 anti-IL-6 antibody were successful in suppressing transaminase elevation for as long as 20 h after Con A injection. We also tried to further improve the protective effect by using a combination of different monoclonal antibodies for the two injections: 20F3 followed by 6B4. As can be seen in Table 4, this combination did indeed suppress transaminase elevation but rather less effectively than two injections of 20F3 antibody. Prompted by the observation that IL-6-deficient mice are more sensitive to Con A-induced hepatitis, we also tested the effect of an antibody treatment regimen aimed at neutralizing not only early Con A-induced IL-6 but also any base-line IL-6 present before Con A challenge. Anti-IL-6 antibody was given by two injections, the first one intraperitoneally 24 h before the Con A challenge, the second one intravenously 1 h before Con A. The rationale for this design was that we relied on the intraperitoneal dose to neutralize base-line IL-6 for a 24-h period and on the intravenous dose to neutralize immediate early Con A-induced IL-6. Blood samples for IL-6 and ALT assays were taken at different time points (Fig. 3). The antibody treatment again had a biphasic effect on levels of biologically active IL-6, an early period of neutralization of base-line as well as Con A-induced IL-6, being followed by 3to 10-fold augmentation at later hours (Fig. 3A). This early blockage of IL-6 activity was accompanied by more pronounced liver damage as evident from ALT levels (Fig. 3B). This increase in liver damage after pretreatment with anti-IL-6 antibody contrasted with transient protection provided by treatment after Con A challenge, but corresponded with increased sensitivity to Con A hepatitis of IL-6-deficient mice. Thus, it is clear that endogenous IL-6 produced after Con A induction plays a dual role: immediate early IL-6 exerts a protective effect, whereas IL-6 produced at later times augments liver damage. http://www.jleukbio.org

TABLE 3.

Serum Transaminase and IL-6 Levels in BALB/c Mice Given a Single Intravenous Injection of Con Aa and Treated with a Single Injection of Anti-IL-6 Antibody at Different Times

Treatment Antibody

Anti-IL-6 Control IgG Anti-IL-6 Control IgG Anti-IL-6 Control IgG Anti-IL-6 Control IgG Anti-IL-6 Control IgG Anti-IL-6 Control IgG Anti-IL-6 Control IgG Anti-IL-6 Control IgG

Serum levelsb of:

Blood samples Time (h post Con A)

Time (h post Con A)

0

5

0

20

2.5

5

2.5

5

2.5

5

2.5

20

5

7

5

20

N

AST (U/L)

ALT (U/L)

9 10 9 10

953 ⫾ 335 (ns) 3,074 ⫾ 1,089 9,996 ⫾ 2,702 (ns) 7,689 ⫾ 2,722

2,223 ⫾ 1,080 (ns) 7,406 ⫾ 2,507 11,953 ⫾ 2,890 (ns) 13,486 ⫾ 5,296

7 7 5 7 10 10 9 5

184 ⫾ 56** 903 ⫾ 804 790 ⫾ 449* 1,767 ⫾ 1,134 1,295 ⫾ 300** 2,689 ⫾ 517 11,670 ⫾ 1,563 (ns) 11,818 ⫾ 891

113 ⫾ 80* 1,266 ⫾ 1,472 1,258 ⫾ 1,035* 3,740 ⫾ 2,739 1,866 ⫾ 428** 4,307 ⫾ 965 6,377 ⫾ 840 (ns) 6,874 ⫾ 860

5 8 5 8

6,514 ⫾ 1328 (ns) 6,958 ⫾ 753 17,642 ⫾ 3,156 (ns) 14,175 ⫾ 1,052

20,062 ⫾ 4249 (ns) 22,698 ⫾ 1,893 21,690 ⫾ 5,104 (ns) 24,344 ⫾ 2,756

IL-6 (log10 U/mL)

2.82 ⫾ 0.14*** 3.73 ⫾ 0.18

a

Con A, 20 µg/g body weight, i.v.; antibodies or IgG, 0.5 mg/mouse, i.v. Mean ⫾ SE. Statistical analysis of difference between anti-IL-6- and IgG-treated groups (Student’s t test); (ns), not significantly different from control; ** P ⬍ 0.05, *** P ⬍ 0.025, **** P ⬍ 0.00001. b

Fig. 2. Effect of anti-IL-6 antibody injection on circulating biologically active IL-6 levels and on ALT levels in Con A-challenged mice. Groups of six BALB/c mice received intravenous injections of Con A (30 µg/g body weight) and anti-IL-6 antibody (20F3) or control IgG/PBS (0.5 mg, time 2.5 h). Blood samples were taken at times indicated on the schedule. Statistical analysis (Student’s t test): *P ⬍ 0.05, **P ⬍ 0.0005, ***P ⬍ 0.00025, for comparison of anti-IL-6 antibody-treated with corresponding control mice.


93

TABLE 4.

Serum Transaminase Levels in BALB/c Mice Given a Single Intravenous Injection of Con A, and Treated with Two Injections of Anti-IL-6 Antibodies AST (U/L)

Experiment number

ALT (U/L)

Treatmenta

N

8h

20 h

8h

1

Anti-IL-6 Ab (20F3, 20F3) Anti-IL-6 Ab (20F3, 6B4) Control IgGb

5 5 5

618 ⫾ 220* 960 ⫾ 251 2,154 ⫾ 840

1,014 ⫾ 482** 1,876 ⫾ 375* 4,500 ⫾ 1,490

776 ⫾ 397* 1,558 ⫾ 759 4,756 ⫾ 2,029

1,582 ⫾ 759** 2,848 ⫾ 422* 7,268 ⫾ 2,377

2

Anti-IL-6 Ab (20F3, 20F3) Control IgG

6 6

2,913 ⫾ 674 3,208 ⫾ 702

5,252 ⫾ 995* 8,345 ⫾ 963

5,063 ⫾ 1,271 6,662 ⫾ 1,662

3,863 ⫾ 492*** 6,857 ⫾ 826

20 h

a

Antibodies (0.5 mg/mouse) given intravenously in indicated sequence, at 2.5 and 5 h post Con A (20 mg/g body weight). Control IgG and PBS were injected for first and second control, respectively. c Student’s t test for difference with control IgG group, * P ⬍ 0.05, ** P ⬍ 0.025, *** P ⬍ 0.01. b

DISCUSSION Our study addressed the question of whether endogenous IL-6 influences the course of Con A-induced hepatitis, and if so, whether it acts as a disease-promoting or as a disease-limiting factor. Exogenous IL-6 has been reported to protect against the disease if administered before Con A challenge, but to augment disease manifestations if administered after Con A [2, 6]. However, the significance of this observation for the net role of endogenous IL-6 has remained unclear.

We followed two complementary approaches to investigate the role of endogenous IL-6: (1) comparison of IL-6-deficient with IL-6-competent mice, and (2) neutralization of endogenous IL-6 with antibodies. We found that IL-6-deficient mice were dramatically more sensitive to Con A-induced hepatitis, indicating that hepatoprotective pathways triggered by IL-6 predominate over possible hepatotoxic ones. IL-6 is known to induce production by hepatocytes of acute-phase reactants, some of which have been shown to have the ability to protect the liver against cytokine-mediated damage [12, 13]. It therefore seems

Fig. 3. Effect of anti-IL-6 antibody injection on circulating biologically active IL-6 levels and on ALT levels in Con A-challenged mice. Groups of 5–12 BALB/c mice received anti-IL-6 antibody (20F3) in two injections (1.3 mg intraperitoneally and 0.5 mg intravenously) before Con A challenge, as indicated on the time schedule. Blood samples were taken for determination of IL-6 and ALT levels. Statistical analysis (Student’s t test): *P ⬍ 0.05, **P ⬍ 0.025, ***P ⬍ 0.005, ****P ⬍ 0.001, for comparison of anti-IL-6 antibody-treated with corresponding control mice.

94



reasonable to assume that IL-6-deficient mice are more sensitive because they lack the stimulus for an important protective pathway. On the other hand, our studies using neutralizing anti-IL-6 antibodies revealed a more complex role of endogenous IL-6. For this analysis we could take advantage of the special property of these antibodies to neutralize endogenous IL-6 for only short time periods, followed by accumulation of IL-6 in the form of antigen-antibody complexes. From studies involving the use of anti-IL-6 antibodies in animals and in humans [17, 20, 22], it has appeared that injected antibodies are initially in excess over available IL-6 and hence neutralize all or most of the biological activity. Then, in a second phase, as the proportion of IL-6 in the accumulating complexes increases, biological activity increases to reach levels superior to those in control animals that did not receive the antibody. Similarly, in this study we found that injection of the anti-IL-6 antibody before or after Con A resulted in transiently lower levels of circulating biologically active IL-6, followed by a phase of dramatically increased levels. Significantly, the effect of the antibody injections on transaminase release depended on the time of antibody treatment. Injection of anti-IL-6 antibody at 2.5 h after the Con A challenge transiently suppressed both IL-6 levels and elevation of serum transaminases. Both effects could be prolonged by adding a second antibody injection at 5 h after Con A challenge. By contrast, anti-IL-6 antibody treatment, initiated well before Con A injection, caused accelerated transaminase elevation. Injection of antibody at an intermediate time point, namely simultaneously with the Con A challenge, had no significant effect, probably because protective and disease-enhancing effects counterbalanced each other. The protective effect of late antibody treatment is indicative of the ability of late endogenous IL-6 to be harmful for the Con A-exposed liver. On the other hand, the acceleration of disease by neutralization of only early IL-6 is indicative of this phase of IL-6 production to be critical for activation of a hepatoprotective mechanism, as was also suggested by the comparison of IL-6-deficient and wild-type mice and by the observation of Mizuhara et al. [2] that administration of IL-6 before Con A induction exerts a protective effect against liver damage. A similar protective effect of pre- but not post-treatment with IL-6 has also been reported in mice given a lethal challenge with endotoxin [23]. In previous studies we and others have demonstrated the importance of IFN-␥, Fas/FasL, and TNF-␣ as diseasepromoting cytokines in the Con A-induced hepatitis syndrome [3–11]. Our present investigation of the role of IL-6 was instigated by the observation that IFN-␥-deficient mice, which are relatively resistant to the syndrome, produce lower splenic IL-6 mRNA and lower circulating IL-6 levels than IFN-␥competent mice. This raised the question of whether the disease-promoting effect of IFN-␥ might somehow be mediated by enhanced IL-6 production. The double-edged role of IL-6 in the syndrome, as demonstrated by the present investigation, makes it difficult to prove or disprove this possibility. However, from our data (Table 1) it appears that enhancement of IL-6 production by IFN-␥ concerns the late rather than the early phases of the Con A syndrome. Because late IL-6 exerts a

disease-promoting effect, enhanced IL-6 production by IFN-␥ can in part explain the difference in Con A sensitivity between IFN-␥-deficient and IFN-␥-competent mice. Gene knock-out mice have proven to be an extremely powerful tool to analyze the role of cytokines in animal models for disease. In most cases the results from experiments with gene knockout mice have confirmed and extended insights gained from experiments using neutralizing antibodies. Our experiments on the role of IL-6 in Con A-induced hepatitis show that each approach has its limitation, and that their combined use is needed to unravel complex systems.

ACKNOWLEDGMENTS Financial support for this work was received from the Belgian Federal Government (IUAP initiative), the Government of Flanders (GOA initiative), and the Fund for Scientific Research (FWO) Flanders. Y. Tagawa was recipient of a 97/98 FWO postdoctoral fellowship. The authors gratefully acknowledge the advice and technical support of Mr. Shigeru Kakuta (Institute of Medical Science, University of Tokyo), as well as the helpful suggestions of Prof. J. Van Damme (Rega Institute). The technical assitance of R. Conings is appreciated.

REFERENCES 1. Tiegs, G., Hentschel, J., Wendel, A. (1992) A T cell-dependent experimental liver injury in mice inducible by concanavalin A. J. Clin. Invest. 90, 196–203. 2. Mizuhara, H., O’Neill, E., Seki, N., Ogawa, T., Kusunoki, C., Otsuka, K., Satoh, S., Niwa, M., Senoh, H., Fujiwara, H. (1994) T cell activationassociated hepatic injury: mediation by tumor necrosis factors and protection by interleukin-6. J. Exp. Med. 179, 1529–1537. 3. Gantner, F., Leist, M., Kusters, S., Vogt, K., Volk, H. D., Tiegs, G. (1996) T cell stimulus-induced crosstalk between lymphocytes and liver macrophages results in augmented cytokine release. Exp. Cell. Res. 229, 137–146. 4. Gantner, F., Leist, M., Lohse, A. W., Germann, P. G., Tiegs, G. (1995) Concanavalin A-induced T-cell-mediated hepatic injury in mice: the role of tumor necrosis factor. Hepatology 21, 190–198. 5. Kusters, S., Gantner, F., Kunstle, G., Tiegs, G. (1996) Interferon-␥ plays a critical role in T cell-dependent liver injury in mice initiated by concanavalin A. Gastroenterol. 111, 462–471. 6. Mizuhara, H., Uno, M., Seki, N., Yamashita, M., Yamaoka, M., Ogawa, T., Kaneda, K., Fujii, T., Senoh, H., Fujiwara, H. (1996) Critical involvement of interferon-␥ in the pathogenesis of T-cell activation-associated hepatitis and regulatory mechanisms of interleukin-6 for the manifestations of hepatitis. Hepatol. 23, 1608–1615. 7. Tagawa, Y., Sekikawa, K., Iwakura, Y. (1997) Suppression of concanavalin A-induced hepatitis in IFN-g-/- mice, but not in TNF-␣-/- mice. J. Immunol. 159, 1418–1428. 8. Louis, H., Le Moine, O., Peny, M. O., Quertinmont, E., Fokan, D., Goldman, M., Deviere, J. (1997) Production and role of interleukin-10 in concanavalin A-induced hepatitis in mice. Hepatol. 25, 1382–1389. 9. Seino, K., Kayagaki, N., Takeda, K., Fukao, K., Okumura, K., Yagita, H. (1997) Contribution of Fas ligand to T cell-mediated hepatic injury in mice. Gastroenterol. 112, 1315–1322. 10. Ksontini, R., Colagiovanni, D. B., Josephs, M. D., Edwards, C. K. R., Tannahill, C. L., Solorzano, C. C., Norman, J., Denham, W., Salzler, C., MacKay, S. L., Moldawer, L. L. (1998) Disparate roles for TNF-␣ and Fas ligand in concanavalin A-induced hepatitis. J. Immunol. 160, 4082–4089. 11. Tagawa, Y., Kakuta, S., Iwakura, Y. (1998) Involvement of Fas/Fas ligand system-mediated apoptosis in the development of concanavalin A-induced hepatitis. Eur. J. Immunol. 28, 4105–4113. 12. Libert, C., Brouckaert, P., Fiers, W. (1994) Protection by ␣1-acid glycoprotein against tumor necrosis factor-induced lethality. J. Exp. Med. 180, 1571–1575.


95

13. Libert, C., Van Molle, W., Brouckaert, P., Fiers, W. (1996) ␣1-antitrypsin inhibits the lethal response to TNF in mice. J. Immunol. 157, 5126–5129. 14. Kopf, M., Baumann, H., Freer, G., Freudenberg, M., Lamers, M., Kishimoto, T., Zinkernagel, R., Bluethmann, H., Kohler, G. (1994) Impaired immune and acute-phase responses in interleukin-6-deficient mice. Nature 368, 339–342. 15. Vink, A., Coulie, P., Warnier, G., Renauld, J. C., Stevens, M., Donckers, D., Van Snick, J. (1990) Mouse plasmacytoma growth in vivo: enhancement by interleukin 6 (IL-6) and inhibition by antibodies directed against IL-6 or its receptor. J. Exp. Med. 172, 997–1000. 16. Starnes, H. F., Jr., Pearce, M. K., Tewari, A., Yim, J. H., Zou, J. C., Abrams, J. S. (1990) Anti-IL-6 monoclonal antibodies protect against lethal Escherichia coli infection and lethal tumor necrosis factor-␣ challenge in mice. J. Immunol. 145, 4185–4191. 17. Heremans, H., Dillen, C., Put, W., Van Damme, J., Billiau, A. (1992) Protective effect of anti-interleukin (IL)-6 antibody against endotoxin, associated with paradoxically increased IL-6 levels. Eur. J. Immunol. 22, 2395–2401. 18. Chomczynski, P., Sacchi, N. (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162, 156–159. 19. Chiu, C. P., Moulds, C., Coffman, R. L., Rennick, D., Lee, F. (1988)

96


20.

21.

22.

23.

Multiple biological activities are expressed by a mouse interleukin-6 cDNA clone isolated from bone marrow stromal cells. Proc. Natl. Acad. Sci. USA 85, 7099–7103. Martens, E., Dillen, C., Put, W., Heremans, H., Van Damme, J., Billiau, A. (1993) Increased circulating interleukin-6 (IL-6) activity in endotoxin challenged mice pretreated with anti-IL-6 antibody is due to IL-6 accumulated in antigen-antibody complexes. Eur. J. Immunol. 23, 2026– 2029. Matthys, P., Froyen, G., Huang, S., Sobis, H., Van Damme, J., Aguet, M., Billiau, A. (1995) Interferon-␥ receptor-deficient mice are hypersensitive to the anti-CD3-induced cytokine release syndrome and thymocyte apoptosis: Protective role of endogenous nitric oxide. J. Immunol. 155, 3823–3829. Lu, Z. Y., Brochier, J., Wijdenes, J., Brailly, H., Bataille, R., Klein, B. (1992) High amounts of circulating interleukin (IL)-6 in the form of monomeric immune complexes during anti-IL-6 therapy. Towards a new methodology for measuring overall cytokine production in human in vivo. Eur. J. Immunol. 22, 2819–2824. Yoshizawa, K. I., Naruto, M., Ida, N. (1996) Injection time of interleukin-6 determines fatal outcome in experimental endotoxin shock. J. Interferon Cytokine Res. 16, 995–1000.
