Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

T lymphocytes are causally involved as effector cells evoking liver disease caused by viral infection (i.e., viral hepatitis) or autoimmune disorders such as chronic active hepatitis, autoimmune hepatitis, primary biliary cirrhosis or primary sclerosing cholangitis (reviewed in Meyer zum Büschenfelde et al., 1993). As a consequence of the activation of these cells, proinflammatory cytokines are systemically released, which possibly accounts for the pathophysiological outcome and clinical manifestation of liver disease. The most potent proinflammatory cytokine in this context is TNF. TNF causes hepatic failure in humans (for review, see Jones and Selby, 1989; Waage, 1993), a phenomenon that has been widely studied in various experimental animal models, mostly in rodents. Pretreatment of mice with inhibitors of hepatic transcription such as GalN renders them extremely sensitive towards TNF-inducible liver failure (Tiegs et al., 1989, 1990; Wendel, 1990) and lethality (Galanos et al., 1979; Lehmann et al., 1987). In these experimental models either TNF itself, or more commonly, LPS, a potent inducer of endogenous TNF production by activation of monocytes and macrophages, is injected into mice. The use of superantigens such as SEB (Miethke et al., 1992; Nagaki et al., 1994; Gantner et al., 1995a) or of agonistic T cell receptor antibodies (Gantner et al., 1995a) as T cell activators in GalN-sensitized animals or of the T cell mitogen Con A in nonsensitized mice (Tiegs et al., 1992) expanded these experimental settings to a variety of different T cell-specific models. In any of them, TNF has been identified as the distal hepatotoxic mediator because passive immunization against TNF completely protected mice from liver injury independent of the T cell activator used (Miethke et al., 1992; Nagaki et al., 1994; Mizuhara et al., 1994; Gantner et al., 1995a,b).

These facts suggest that TNF suppression might be a promising pharmacological approach to prevent liver damage induced by T cell activation. Cyclic AMP-elevating drugs such as PDE inhibitors are known to inhibit monocyte- and macrophage-derived TNF production after LPS stimulation in vitro (Semmler et al., 1993; Fischer et al., 1993; Reinstein et al., 1994; Prabhakar et al., 1994; Seldon et al., 1995; Jilg et al., 1996) and in vivo (Fischer et al., 1993; Jilg et al., 1996). In human peripheral blood mononuclear cell cultures (Essayan et al., 1994; Gantner et al., 1995c) or in the murine T_H2 cell line D10G4.1 (Schmidt et al., 1995), the combined inhibition of PDE3 and PDE4 was most effective in blocking T cell proliferation and in modulating cytokine production. Recent reports gave a first hint that PDE4-selective inhibitors such as the antidepressant rolipram might be beneficial in vivo in chronic T cell-mediated disorders such as experimental allergic encephalomyelitis, thought to be a model for multiple sclerosis, in which TNF also plays a significant role (Sommer et al., 1995; Genain et al., 1995). We investigated the in vivo potential of PDE inhibitors selective for PDE3 (motapizone) or PDE4 (rolipram), respectively, in our acute models of T cell-mediated hepatic failure. With these inhibitors, we studied their modulatory effects on the T cell activation-induced cytokine pattern released and on the outcome of liver failure after T cell activation in naive mice (Con A model) and in animals sensitized by transcriptional inhibition (GalN/SEB model).

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Animals. Male BALB/c mice weighing 25 ± 3 g were purchased from the internal breeding stock of the animal house of University of Konstanz, Germany. All mice were kept at 22°C and 55% relative humidity in a 12-h day/night rhythm with free access to food (Altromin 1313) and water. Sixteen hours before the experiment started food was withdrawn, and the mice were not fed overnight. All animals received humane care in compliance with National Institutes of Health guidelines and according to legal requirements in Germany.

Treatment schedules. Con A was purchased from Sigma Chemical Co. (St. Louis, MO) and 25 mg/kg were administered i.v. in a volume of 300 µl pyrogen-free saline into the tail vein. GalN (Roth Chemicals, Karlsruhe, Germany) was given intraperitoneally at a dose of 700 mg/kg in 200 µl saline 15 min before injection of SEB (Sigma; 2 mg/kg i.p. in 200 µl saline). Rolipram (4-(3-cyclopentyloxy-4-methoxyphenyl)-2-pyrrolidone) and motapizone (4,5-dihydro-6-[4-(11-imidazol-1-yl)-2-thienyl]-5-methyl-3-pyridazinone) were administered i.p. 30 min before challenge in a volume of 200 µl solvent (Sandimmune placebo, Sandoz, Basel, Switzerland). The same volume of solvent solution was injected into control animals. For determination of time courses of cytokine release into the circulation, animals were sacrificed at various time points after challenge by injection of heparinized pentobarbital (150 mg/kg), and blood was withdrawn by heart puncture into heparinized syringes. Blood sampling from the tail vein for determination of plasma cytokine peak concentrations was performed at the time points indicated. Plasma samples were kept frozen at 70°C until further use.

Cytokine assays. All enzyme-linked immunosorbent assays were performed on flat-bottomed, high-binding polystyrene microtiter plates (Greiner, Nürtingen, Germany) with use of specific rat anti-mouse monoclonal antibody pairs purchased from Pharmingen (San Diego, CA). A polyclonal monospecific ovine anti-mouse TNF antibody prepared by S. Jilg in the laboratory of Dr. A. Wendel (immunoglobulin G fraction; protein content, 20 mg/ml) after immunization of a ram with recombinant muTNF- was used for coating instead of the Pharmingen antibody. Streptavidin-peroxidase was from Jackson Immuno Research (West Grove, PA) and the peroxidase substrate BM blue was from Boehringer (Mannheim, Germany). IL-10 was determined with the INTERTEST-10×TM ELISA kit (Genzyme, Cambridge, MA). The detection limits of the assays were 15 pg/ml for IL-10, 10 pg/ml for TNF and IFN and 5 pg/ml for IL-4, respectively.

Assessment of liver failure. Eight hours after Con A or SEB injection animals were sacrificed by cervical dislocation, and blood was collected by heart puncture. Plasma enzyme activities of ALT, AST and SDH, respectively, were assessed according to Horder and Rej (1984).

Statistics. Unless otherwise stated, data are expressed as means ± S.D. of experiments carried out in triplicate. Data were analyzed by nonparametric analysis of variance (Kruskal-Wallis), and where there were differences among the groups (P > .05) data were subjected to one-sided nonparametric multiple comparisons of the disease control group against all other groups (Zar, 1984). P < .05 was considered to be significant.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

T cell activation-induced cytokine release in vivo. In previous studies we have shown the plasma kinetics of the cytokines TNF, IFN, IL-2 and GM-colony-stimulating factor after Con A challenge of mice (Gantner et al., 1995b). Because PDE inhibitors were recently demonstrated to affect murine T_H2-type cytokine production in vitro (Schmidt et al., 1995), we first checked whether and when the T_H2-type cytokines IL-4 and IL-10 were released in our in vivo models. The T cell activation-induced IL-4 and IL-10 plasma cytokine levels are shown in figure 1 in comparison with the time course of TNF release, i.e., the central mediator of Con A- and GalN/SEB-induced liver failure, and in comparison with the T_H1-type cytokine IFN. Con A induced detectable amounts of IL-4 as early as 30 min after challenge with a further increase up to a maximum concentration of about 900 pg/ml at 2 to 2.5 h that declined afterward. IL-10 was released in a biphasic manner characterized by a high first peak at 30 min (4600 ± 420 pg/ml). This was followed by complete absence of circulating IL-10 at 4 h, before a second phase of IL-10 production was noted 6 to 8 h after Con A challenge.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Time course of cytokine release into plasma of BALB/c mice injected i.v. with 25 mg/kg Con A (filled symbols) or GalN (700 mg/kg i.p.)/SEB (2 mg/kg i.p., open symbols). Data are expressed as mean values from a single experiment with three mice (IL-4) or as mean values from two independent experiments using three animals each (n = 6; TNF, IFN, IL-10) ± S.D. At the time points indicated mice were sacrificed and the plasma samples taken were frozen at 70°C until cytokine determination by enzyme-linked immunosorbent assay. The maximum circulating cytokine concentrations (in pg/ml) were as follows: for Con A: TNF, 1120 ± 160; IFN, 4130 ± 110; IL-4, 930 ± 120; IL-10, 4460 ± 420; for SEB: TNF, 360 ± 120; IFN, 19,200 ± 8,160; IL-4, 220 ± 120; IL-10, 6470 ± 1870. None of the four cytokines was detectable in mice treated with pyrogen-free saline or GalN alone (700 mg/kg i.p.).

Modulation of the cytokine pattern by PDE inhibition. We then examined the pharmacological consequences of PDE inhibition on cytokine production after T cell activation in vivo. For this purpose we treated mice with various doses of either motapizone, a PDE3 inhibitor, or rolipram, a type 4-selective PDE inhibitor, before injection of Con A or GalN/SEB, respectively, and determined the circulating levels of TNF, IL-4 and IL-10 2 h after Con A and 2.5 h after GalN/SEB administration. IFN was determined 8 h after challenge, i.e., at the end of the experiment. These time points were chosen, because then all cytokines of interest were present in considerable amounts in the circulation of the animals and could be determined out of one single sample of each individual animal.

View larger version (33K): [in this window] [in a new window]   Fig. 2.   Modulation by motapizone of the cytokine pattern induced by Con A (25 mg/kg i.v.) in nonsensitized (A) or SEB (2 mg/kg i.p.) in GalN (700 mg/kg i.p.)-sensitized (B) mice. Motapizone was administered i.p. 30 min before challenge at the doses indicated. The corresponding volume of solvent was injected into control animals. Blood samples for the determination of TNF, IL-4 and IL-10 were taken from the tail vein at 2 h after Con A or 2.5 h after SEB injection. IFN concentrations were determined 8 h after challenge. Data are mean values ± S.D. (n = 3 mice per group) from one out of two analogous experiments. *P < .05 versus solvent control.

View larger version (28K): [in this window] [in a new window]   Fig. 3.   Modulation by rolipram of the cytokine pattern induced by Con A (25 mg/kg i.v.) in nonsensitized (A) or SEB (2 mg/kg i.p.) in GalN (700 mg/kg i.p.)-sensitized (B) mice. Rolipram was administered i.p. 30 min before challenge at the doses indicated. The corresponding volume of solvent was injected into control animals. Blood samples for the determination of TNF, IL-4 and IL-10 were taken from the tail vein at 2 h after Con A or 2.5 h after SEB injection. IFN concentrations were determined 8 h after challenge. Data are mean values ± S.D. (n = 3 mice per group) from one out of two independent experiments. *P < .05 versus solvent control.

Prevention of T cell-dependent hepatic failure by PDE inhibition. Based on the observation that PDE inhibition in vivo strongly suppressed the release of the central mediator for development of hepatic failure, i.e., TNF, we tested the hypothesis that motapizone and rolipram might prevent liver failure after injection of hepatotoxic amounts of Con A or GalN/SEB. Injection of Con A (25 mg/kg i.v.) in naive mice or of SEB (2 mg/kg i.p.) in GalN-sensitized animals (700 mg/kg) evoked fulminant liver failure within 8 h as assessed by highly increased plasma activities of liver-specific enzymes ALT and SDH, which indicated massive liver cell destruction (cf. figs. 4 and 5). Pretreatment of such mice with either motapizone (fig. 4) or rolipram (fig. 5) dose-dependently prevented development of liver failure in both models. Plasma ALT activities were indistinguishable from untreated control values (35 ± 15 U/l) at the highest drug dose used, and the behavior of the animals was normal. It is important to point out that the dose-response relation of either compound with regard to TNF suppression and protection from liver failure was similar. Thus, protection by the PDE3 inhibitor motapizone as well as by rolipram, a PDE4 inhibitor, was most likely caused by suppression of TNF production.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Protection from T cell-dependent murine liver failure by motapizone. Liver damage was induced by injection of Con A (25 mg/kg i.v.) in nonsensitized (A) or SEB (2 mg/kg i.p.) in GalN (700 mg/kg i.p.)-sensitized (B) mice. Motapizone was administered i.p. 30 min before challenge at the doses indicated. The corresponding volume of solvent was injected into control animals. Eight hours after challenge, animals were sacrificed, and plasma enzyme activities of ALT, AST and SDH were determined. Data are mean values ± S.D. of n = 3 animals from a single experiment (0.01 mg/kg motapizone) or of two independent experiments using three animals each (n = 6 for all other groups). *P < .05 versus solvent control.

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Protection from T cell-dependent murine liver failure by rolipram. Liver damage was induced by injection of Con A (25 mg/kg i.v.) in nonsensitized (A) or SEB (2 mg/kg i.p.) in GalN (700 mg/kg i.p.)-sensitized (B) mice. Rolipram was administered i.p. 30 min before challenge at the doses indicated. The corresponding volume of solvent was injected into control animals. Eight hours after challenge, animals were sacrificed, and plasma enzyme activities of ALT, AST and SDH were determined. Data are mean values ± S.D. of n = 3 animals from a single experiment (0.01 mg/kg rolipram) or of two independent experiments using three animals each (n = 6 for all other groups). *P < .05 versus solvent control.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

This study investigated the pharmacological influence of PDE inhibitors on T cell activation-induced, cytokine-dependent liver injury in mice. We showed in two different murine models of fulminant hepatic failure, i.e., the Con A model and the GalN/SEB model, that motapizone, a PDE3 inhibitor, as well as rolipram, a PDE4-selective inhibitor, shifted the T cell-dependent cytokine response from proinflammatory cytokine production toward increased release of the anti-inflammatory cytokine IL-10. We conclude that this change in cytokine response conferred protection of mice from T cell-dependent liver damage.

There are two principles by which PDE inhibitors could have exerted their protective effects: 1) either by acting on the effector cell level, i.e., on the cell population(s) producing liver-toxic cytokines such as TNF and IFN, or 2) by acting on the target cell level, i.e., by interfering with the toxic signal(s) leading to hepatocyte destruction, or 3) by both principles. Indeed, it has been shown that rolipram protects mice not only from GalN/LPS-induced liver injury by reduction of TNF production but also from GalN/TNF-induced hepatic failure (Fischer, et al., 1993). This observation is in accordance with recent in vitro results demonstrating cytoprotection in murine hepatocyte cultures by mixed PDE inhibitors against actinomycin D/TNF exposure (Jilg et al., 1996).

With respect to this study the reduced circulating TNF levels measured in PDE-inhibitor-pretreated animals are sufficient to explain the protective effect of these drugs against T cell-dependent hepatic failure. Protection from Con A might have been potentiated by simultaneous inhibition of IFN generation, a lymphokine that has been identified recently as an additional critical mediator in Con A-induced liver injury (Küsters et al., 1996). Our results are in line with the known capacity of PDE4 inhibitors to suppress LPS-induced TNF production (Zabel et al., 1993; Semmler et al., 1993; Schade and Schudt, 1993; Fischer et al., 1993; Molnar-Kimber et al., 1993; Prabahkar et al., 1994; Seldon et al., 1995; Leist et al., 1996). Most likely, elevated cAMP levels account for the mechanism of action of these drugs. This nucleotide mediator acts via a PKA pathway that finally can result in negative (IL-2) or positive (IL-6, IL-10) regulation of cytokine gene expression via a so-called CRE sequence on the promoter region (Novak and Rothenberg, 1990; Munoz et al., 1990; Dendorfer et al., 1994; Platzer et al., 1995). The mechanism leading to cAMP-dependent inhibition of TNF production is less clear, because a CRE consensus sequence has been identified neither on the murine nor on the human TNF gene promoter enhancer region. Moreover, rolipram failed to influence TNF mRNA expression in murine peritoneal macrophages stimulated with endotoxin (Kambayashi et al., 1995). Alternatively, cAMP-dependent regulation of TNF production at the posttranscriptional level has been suggested (Severn et al., 1992). Recently, an indirect mode of cAMP-regulated cytokine synthesis involving the PKA pathway has been demonstrated for IL-5 (Lee et al., 1993). Another explanation might be a cAMP/PKA-mediated modulation of the activity of nuclear factors. In activated T cells, cAMP was shown to decrease NF-AT induction and that consequently leads to up-regulation of IL-4 and IL-5 (Lacour et al., 1994). Inhibition by cAMP of NF-B activation, a molecular event necessary for the induction of the TNF gene expression in B and T cells (Goldfeld et al., 1994), could represent an alternate mechanistic explanation of diminished TNF production observed after PDE inhibition.

In addition to a possible NF-B effect, two more TNF-regulatory principles should be taken into account: first, the reduced TNF plasma levels noted on PDE inhibition could be the consequence of increased TNF receptor shedding after cAMP elevation. Such a mechanism is believed to account for the protective effect of methylxanthines in murine models of septic liver failure (Jilg et al., 1996). Second, recent reports as well as our present study make endogenous IL-10 a likely candidate to contribute to the sequence of events finally resulting in low plasma concentrations of TNF. IL-10 is known to suppress TNF production in a variety of experimental settings and IL-10 neutralization leads to elevation of LPS-induced TNF (Fiorentino et al., 1991; Bogdan et al., 1991; reviewed in Moore et al., 1993). Moreover, it has been shown that endogenously produced IL-10 (Florquin et al., 1994) or exogenously administered rmuIL-10 (Bean et al., 1993) counteract the toxic effects of TNF induced by GalN/SEB challenge of mice. Sequence analysis studies have demonstrated a CRE on both the murine and human IL-10 promoter region (Kim et al., 1992; Platzer et al., 1995), which indicates cAMP regulation of IL-10 synthesis. Indeed, cAMP-elevating drugs such as mixed-type PDE inhibitors or adenylyl cyclase stimulants as well as direct administration of dibutyryl cAMP have been reported to increase LPS-induced IL-10 production (Platzer et al., 1995; Jilg et al., 1996; Arai et al., 1995). Moreover, anti-TNF antibodies have been shown to result in an increase of LPS-induced IL-10 production in mice (Barsig et al., 1995). Regarding the IL-10 plasma kinetics after in vivo T cell activation in comparison with plasma TNF levels (cf. fig. 1), a role of IL-10 in down-regulation of TNF in our T cell-dependent models seems possible.

Finally, the question arises of whether motapizone and rolipram, respectively, at their protective doses used, still affected PDE3 or PDE4 isozymes selectively. On the basis of our present in vivo data, this question remains open. However, the fact that zaprinast, a PDE5-selective compound, did not protect mice from GalN/SEB-induced liver failure (data not shown), makes a critical regulatory contribution of PDE families other than PDE3 and PDE4 to T cell cytokine synthesis rather unlikely. This assumption is further supported by the complete protection from GalN/SEB-initiated liver failure that was observed by the combined use of both compounds (cf. table 1) at doses at which these drugs were ineffective with regard to plasma liver enzyme activities when given alone. Whether motapizone and rolipram in vivo act in an additive or in a synergistic manner is difficult to answer for several reasons. First, although these drugs are selective for PDE3 or PDE4, respectively, they loose their selectivity at higher concentrations and might have affected both isoenzymes, probably even at the dose of 0.1 mg/kg. Second, regarding modulation of circulating cytokines, the combined use of motapizone and rolipram was additive, whereas ALT levels were reduced synergistically. This apparent synergism with regard to protection from liver injury might be the consequence of an additive effect on TNF reduction (cf. table 1).

However, our observations argue in favor of an important role of both isozymes for the regulation of T cell-mediated inflammatory processes. Accordingly, T cells mainly express PDE3 and PDE4 isozymes (Tenor et al., 1995) and from several in vitro studies in human (Essayan et al., 1994; Gantner et al., 1995c) as well as murine T cells (Schmidt et al., 1995), a prominent role of both PDE3 and PDE4 in regulation of T cell activation became evident. Thus, our data seem to translate these in vitro findings to the in vivo situation by demonstrating that PDE3 and PDE4 inhibitors potently protect from fulminant T cell-dependent liver failure in mice. We ascribe this protection to a shift of the cytokine response from a proinflammatory to an anti-inflammatory profile. In this sense, our findings corroborate the concept that cAMP elevation results in suppression of T_H1-type cytokines, whereas T_H2-cytokines are up-regulated (reviewed in Haraguchi et al., 1995). However, the production of the specific T_H2-type cytokine IL-4 was not affected by PDE3 inhibition but was significantly suppressed by rolipram. In contrast to IL-4, IL-10 is not a T_H2-specific cytokine and is also produced by monocytes and macrophages. We recently reported that upon stimulation of macrophages by LPS in vivo the liver and the spleen are the two major IL-10-producing organs of the mouse (Barsig et al., 1995). Hence, it seems likely that the accessory cells of T cell activation in our models, i.e., the macrophages, were rather affected by PDE inhibitors to release increased amounts of IL-10 than were the T_H2-lymphocytes.

Because an increase of T_H2 lymphokines such as IL-4 and IL-5 may evoke complications in patients with allergic disease, our results which showed a selective increase of IL-10 seem to be of considerable interest with respect to the therapeutic use of PDE inhibitors for the treatment of T cell-related disorders.