Superoxide produced by Kupffer cells is an essential effector in concanavalin A–induced hepatitis in mice

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Although concanavalin A (Con-A)-induced experimental hepatitis is thought to be induced by activated T cells, natural killer T (NKT) cells, and cytokines, precise mechanisms are still unknown. In the current study, we investigated the roles of Kupffer cells, NKT cells, FasL, tumor necrosis factor (TNF), and superoxide in Con-A hepatitis in C57BL/6 mice. Removal of Kupffer cells using gadolinium chloride (GdCl3) from the liver completely inhibited Con-A hepatitis, whereas increased serum TNF and IFN-γ levels were not inhibited at all. Unexpectedly, anti-FasL antibody pretreatment did not inhibit Con-A hepatitis, whereas it inhibited hepatic injury induced by a synthetic ligand of NKT cells, α-galactosylceramide. Furthermore, GdCl3 pretreatment changed neither the activation-induced down-regulation of NK1.1 antigens as well as T cell receptors of NKT cells nor the increased expression of the CD69 activation antigen of hepatic T cells. CD68+ Kupffer cells greatly increased in proportion in the early phase after Con-A injection; this increase was abrogated by GdCl3 pretreatment. Anti-TNF antibody (Ab) pretreatment did not inhibit the increase of Kupffer cells, but it effectively suppressed superoxide/reactive oxygen production from Kupffer cells and the resulting hepatic injury. Conversely, depletion of NKT cells in mice by NK1.1 Ab pretreatment did suppress both the increase of CD68+ Kupffer cells and Con-A hepatitis. Consistently, the diminution of oxygen radicals produced by Kupffer cells by use of free radical scavengers greatly inhibited Con-A hepatitis without suppressing cytokine production. However, adoptive transfer experiments also indicate that a close interaction/cooperation of Kupffer cells with NKT cells is essential for Con-A hepatitis. Conclusion: Superoxide produced by Kupffer cells may be the essential effector in Con-A hepatitis, and TNF and NKT cells support their activation and superoxide production. (HEPATOLOGY 2008;48:1979-1988.)

It is well known that severe hepatic injury is induced after Concanavalin A (Con-A) administration into mice. This experimental hepatitis model, which was discovered by Tiegs et al.1 in 1992, is accompanied by a remarkable activation of T cells. Considering the indispensable involvement of CD4-positive T cells, Con-A hepatitis has been regarded as a model of T cell–mediated autoimmune hepatitis.1 Since the discovery of Con-A hepatitis, a number of studies have been published. Among the wide range of cytokines, the critical role of tumor necrosis factor (TNF) in Con-A hepatitis was proved by experiments using anti-TNF antibody (Ab).2–4 It was also reported that interferon gamma (IFN-γ) participates strongly in the manifestation of hepatotoxicity.5, 6

In previous reports, natural killer T cells (NKT cells) were suggested to play a major role in this hepatitis model.7–9 Considering that this hepatitis does not develop well in FasL-deficient mice, some researchers have suggested that hepatocyte apoptosis induced by interacting Fas on hepatocytes with FasL on the activated NKT cells is the main mechanism of this hepatitis.9, 10 However, other researchers have demonstrated that blockade of FasL did not inhibit Con-A hepatitis,11 suggesting that FasL is not essential in this disease. Furthermore, nuclear factor kappaB and caspase 8, which are essential for apoptosis, are not involved in this hepatitis.12 Conversely, Kupffer cells are certainly involved in Con-A hepatitis, because the specific blockade of the Kupffer cells by GdCl313 or clodronate liposomes14 completely inhibits this disease. Kupffer cells have a strong TNF secretion capability, and it has been reported that the TNF secreted from Kupffer cells participates in the development of this hepatitis.15

Considering these findings, the developmental mechanism is assumed to be as follows16: After the administration of Con-A, activated Kupffer cells secrete TNF, interleukin-12, and interleukin-18, and they activate T cells. T cells (including NKT cells) produce interferon gamma (IFN-γ) and further activate Kupffer cells in a positive feedback loop. TNF and IFN-γ are believed to be involved in various hepatic injuries, and they also may be hepatotoxic and critical for Con-A hepatitis. However, our recent studies have found that GdCl3 pretreatment of mice reduced superoxide production but increased TNF production from liver-adherent macrophages, and that TNF in turn accelerated the proliferation of regenerating hepatocytes after partial hepatectomy,17, 18 demonstrating that TNF is not necessarily hepatotoxic. These findings led us to examine in detail the roles of Kupffer cells, NKT cells, TNF, and oxygen radicals in Con-A hepatitis.

As we show in this study, mice whose Kupffer cells have been removed by gadolinium chloride (GdCl3) treatment do not show hepatitis; however, the production capacities for TNF and IFN-γ do not change at all. We explore that the production of superoxide and reactive oxygen species (ROS) from the Con-A–activated Kupffer cells plays a critical role in this hepatitis.

Abbreviations

α-GalCer, α-galactosylceramide; Ab, antibody; ALT, alanine aminotransferase; Con-A, concanavalin-A; IFN-γ, interferon gamma; lec-SOD, lecithinized superoxide dismutase; MNCs, mononuclear cells; NK, natural killer; NK1.1, natural killer 1.1; NKT cells, natural killer T cells; PBS, phosphate-buffered saline; ROS, reactive oxygen species; SE, standard error; TCR, T cell receptor; TNF, tumor necrosis factor.

Materials and Methods

This study was conducted according to the guidelines of the Institutional Review Board for the Care of Animal Subjects at the National Defense Medical College, Japan.

Mice.

Male C57BL/6 (B6) young (7–8 weeks of age) mice, middle-aged B6 mice (30 weeks of age), and young BALB/c mice were purchased from Japan SLC. Because B6 CD1d−/− mice were not commercially available, CD1d−/− mice with a BALB/c background were purchased from the Jackson Laboratory (USA) and were mated to each other. CD1d−/− mice of 7 to 8 weeks of age were used.

Reagents.

Con-A was purchased from Vector Laboratories, Inc. (Burlingame, CA) and was injected intravenously into young mice (0.25 mg/mouse) via the caudal vein. Phosphate-buffered saline (PBS) was injected as a control. Alpha-galactosylceramide (α-GalCer)19, 20 was provided by the Pharmaceutical Research Laboratory of the Kirin Brewery Company, and middle-aged mice were injected intravenously with α-GalCer at 100 μg/kg body mass. To remove the Kupffer cells, 200 μg per mouse of gadolinium chloride (GdCl3) (Sigma Chemical Co., St. Louis, MO) was injected intravenously via the caudal vein from 24 hours to 36 hours before Con-A administration.

Isolation of Mononuclear Cells.

Hepatic mononuclear cells (MNCs) were prepared essentially as described previously.21 Briefly, liver specimens were minced with scissors. After adding 0.25% collagenase solution, specimens were shaken for 15 minutes in a 37°C constant-temperature bath. Next, the liver specimen was filtered through a stainless steel mesh. After mixing in 33% Percoll solution containing heparin, the sample was centrifuged for 15 minutes at 500g at room temperature. After removing the supernatant, the pellet was resuspended in a red blood cell lysis solution and then was washed twice in 10% fetal bovine serum (FBS) Roswell Park Memorial Institute (RPMI) 1640.

Flow Cytometry.

Phycoerythrin Cy5–conjugated monoclonal antibodies to mouse T cell receptor (TCR) alpha beta (H57–597), and phycoerythrin monoclonal antibody to natural killer 1.1 (NK1.1) (PK136) were obtained from BD PharMingen. To detect invariant Vα14Jα18 NKT cells, α-GalCer–loaded CD1d–immunoglobulin G (IgG) 1 dimers (phycoerythrin-conjugated)22, 23 were used as described in the technical data sheet (BD PharMingen). Before staining with antibodies (Abs), the MNCs were incubated for 10 minutes at 4°C with Fc-blocker (2.4 G2; BD PharMingen) to prevent any nonspecific binding. For identification of Kupffer cells, mononuclear cells were stained by fluorescein isothiocyanate–labeled anti-F4/80 Ab (eBioscience), phycoerythrin-labeled CD11b Ab (PK136, BD PharMingen), and biotin-labeled CD68 Ab (AbD Serotec, Oxford, UK) with Cy5-streptavidin. For the detection of intracellular ROS, aminophenyl fluorescein (Daiichi Chemical Tokyo),24 which has same chemiluminescence as fluorescein isothiocyanate, was added before the addition of staining Ab. Flow cytometry was performed using an Epics XL instrument (Coulter).

Measurement of Alanine Aminotransferase and Cytokine Levels.

After the administration of Con-A, blood was collected from the retro-orbital sinus at specified time points. Serum alanine aminotransferase (ALT) was measured using Drichem 3000V (Fuji Medical Systems, Tokyo). Enzyme-linked immunosorbent assay kits for IFN-γ (BD PharMingen) and for TNF (Endogen) were used.

The Neutralization of TNF, IFN-γ, and FasL, and Removal of Natural Killer Cells and of Both Natural Killer and NKT Cells.

Anti-TNF Abs (0.5 mg/mouse) (MP6-XT3, BD PharMingen), IFN-γ Ab, FasL Ab (0.5 mg/mouse)25 (PharMingen), or PBS was injected intravenously 1 hour before Con-A administration. Anti-asialo GM1 Ab (50 μg/mouse) or anti-NK1.1 Ab (200 μg/mouse) was administered 3 days before Con-A injection to deplete either natural killer (NK) cells alone or both NK cells and NKT cells, respectively.

Depletion of Superoxide and ROS.

To deplete superoxide or ROS in Con-A hepatitis, 200 μg per mouse of 3-methyl-1-phenyl-pyrazolin-5-one (Edaravone) (Mitsubishi Pharmaceutical, Tokyo, Japan)26, 27 was injected intravenously 2 hours before Con-A administration. Lecithinized superoxide dismutase (lec-SOD) was provided by Seikagaku Corporation (Tokyo, Japan),28–30 and 100 μg/mouse lec-SOD was injected 1 hour before Con-A administration. For the control mouse, PBS was injected intravenously.

Determination of Superoxide Production From Kupffer Cells.

Superoxide production of Kupffer cells was determined by 2-methyl-6-phenyl-3,7-dihydroimidazo[31]pyrazin-3-one–dependent chemiluminescence as described.31 In brief, MNCs were suspended in a collagen-coated plate at a concentration of 2.5 × 106/mL in 10% FBS RPMI 1640. After the incubation at 37°C for 2 hours, the culture media was removed. To remove nonadherent cells, plate wells were washed by pipetting with medium maintained at 37°C. Then the culture plate was put on ice and the adherent cells, which include Kupffer cells, were collected by pipetting. Collected cells were resuspended at a concentration of 1.0 × 105/mL in Hanks' balanced salt solution containing 2 μM 2-methyl-6-phenyl-3,7-dihydroimidazo[31]pyrazin-3-one and added to a glass cuvette. Adherent cells were cultured at 37°C while measuring the fluorescence intensity in the cuvette by fluorometry (BLR-101, Aloka, Japan). After 3 minutes, the cell suspension was stimulated by adding 4 μg phorbol 12-myristate 13-acetate, and fluorescence activity was measured. Finally, based on the fluorescence activity after addition of 0.5 μM superoxide dismutase, we calculated the fluorescence by the superoxide secreted by the Kupffer cells.

Irradiation of Mice.

CD1d−/− mice were irradiated (4 Gy) using an MBR-1505R2 (Hitachi Medical Corporation) at 48 hours before adoptive transfer experiments.

Adoptive Transfer Experiments.

Liver MNCs were obtained from BALB/c mice 1 hour after Con-A injection and stained with Cy-5 anti-F4/80 Ab, and F4/80+ Kupffer cells and F4/80 lymphocytes were enriched using the MACS system (Militenyi Biotec, Auburn, CA). The purity of F4/80+ cells was approximately 85%; these cells contain virtually no NKT cells (less than 2%), and F4/80 cells contained few F4/80+ cells (less than 5%). Cells (5 × 106 of either type) were transferred from spleens into BALB/c mice or CD1d−/− mice. Transfers of cells were performed by 30-gauge needle, and the puncture site on the spleen was coagulated by an electric scalpel.

Statistical Analysis.

The data are presented as the mean values ± standard error (SE). Statistical analyses were performed using an iMac computer (Apple, Cupertino, CA) and the Stat View 4.02J software package (Abacus Concepts, Berkeley, CA). Statistical evaluations were compared using the standard one-way analysis of variance followed by the Bonferoni post hoc test. P < 0.05 was considered to indicate a significant difference.

Results

Inhibition of Con-A Hepatitis by Various Reagents.

After Con-A administration to a mouse, serum ALT levels started to rise after approximately 8 hours and peaked between 12 and 24 hours. The most effective way of inhibiting Con-A hepatitis is prior removal of the Kupffer cells by GdCl3 or the neutralization of TNF by anti-TNF Ab (Fig. 1). Anti-IFN-γ Ab mostly abrogated hepatitis (Fig. 1). Neutralizing anti-FasL Ab did not affect elevated serum ALT levels (Fig. 1). Depletion of both NK cells and NKT cells (αNK1.1)—but not depletion of NK cells alone (α Aslalo)—substantially reduced the hepatotoxicity (Fig. 1). In contrast to Con-A hepatitis, hepatic injury induced by a synthetic NKT cell ligand, α-galactosylceramide (α-GalCer), was inhibited by either anti-TNF Ab32, 33 or anti-FasL Ab, but not by either GdCl3 or anti–IFN-γ Ab (Fig. 2). In contrast to α-GalCer–induced hepatic injury,32, 33 age-dependent enhancement of hepatic injury was not observed in Con-A hepatitis (data not shown).

Figure 1.

Inhibition of Con-A hepatitis by pretreatment with several reagents. Young mice were pretreated with the indicated reagents and were injected intravenously with Con-A. Blood samples were obtained from the retroorbital sinus 12 hours after Con-A injection, and ALT levels were measured. Four to six mice in each group were examined. Data are mean ± SE, **P < 0.01.

Figure 2.

Inhibition of hepatic injury induced by α-GalCer by pretreatment of mice with anti-TNF Ab or anti-FasL Ab (but not by GdCl3 nor by anti–IFN-γ Ab). Middle-aged mice were pretreated with the indicated reagents and were injected intravenously with α-GalCer. Blood samples were obtained from the retroorbital sinus 12 hours after α-GalCer injection, and ALT levels were measured. Four to six mice in each group were examined. Data are mean ± SE, **P < 0.01.

Phenotypical Changes of Hepatic T Cells and NKT Cells After Con-A Administration Are Not Altered by GdCl3 Pretreatment.

Activation of NKT cells by either α-GalCer or interleukin-12 reportedly down-regulates TCR or NK1.1 antigen (Ag).34–36 Twelve hours after Con-A administration, either NK1.1 Ag, TCR, or invariant TCR (Vα14Jα18/Vβ8 or Vβ7) expressions were also down-regulated, and CD69 expression (an activation marker) of liver T cells with high TCR increased, whereas GdCl3 pretreatment did not alter these phenotypical changes (Fig. 3). Anti-TNF Ab pretreatment of mice also did not affect these phenotypical changes (data not shown).

Figure 3.

Down-regulation of either NK1.1 Ag, TCR, or invariant TCR of liver NKT cells and increased expression of CD69 of liver T cells with high TCR. Liver MNCs were obtained from mice at 12 hours after Con-A injection and stained with Abs, and gated lymphocytes were analyzed. NKT cells with intermediate TCR are indicated by squares in the upper and middle panels, and data are means ± SE from three mice in each group. Liver T cells with high TCR are also indicated in the lower panels.

Increased Number of CD68-Positive Kupffer Cells After Con-A Administration.

Because the removal of the Kupffer cells by GdCl3 inhibited Con-A–induced hepatotoxicity, we examined the proportion of Kupffer cells after Con-A administration. In control mice that were injected with PBS intravenously, CD68+CD11b Kupffer cells accounted for only approximately 7.5% of total MNCs (Fig. 4). The proportion of CD68+ cells increased up to 24.3% at 1 hour after Con-A administration, and gradually decreased through 3 hours and 6 hours (Fig. 4). These CD68+ cells were also F4/80 positive (data not shown). The total number of liver MNCs obtained from control mice was 3.8 ± 0.5 × 106 (n = 10, mean ± SE), and that of liver MNCs obtained from mice at 1 hour after Con-A injection was 5.0 ± 0.6 × 106 (n = 10, mean ± SE) (P < 0.05).

Figure 4.

Early increase of CD68+ Kupffer cells in mouse liver after Con-A injection. Liver MNCs were obtained from mice at the indicated time points after Con-A injection. Data are representative of five independent experiments, which all had similar results.

Proportional Change of Kupffer Cells in GdCl3− or TNF Ab–Pretreated Mice.

We next examined whether the increase of CD68+ Kupffer cells in the liver is affected by GdCl3 or anti-TNF Ab pretreatment. The proportional increase of Kupffer cells after Con-A injection was lowered from 23.1% to 5.5% in GdCl3-pretreated mice (Fig. 5). However, anti-TNF Ab pretreatment did not decrease the proportion of CD68+ Kupffer cells (Fig. 5).

Figure 5.

Inhibition of the increase of CD68+ Kupffer cells by GadCl3 pretreatment after Con-A injection. Liver MNCs were obtained 1 hour after Con-A injection from normal mice and from mice pretreated with GadCl3 or anti-TNF Ab. Data are representative of five independent experiments, which all had similar results.

Superoxide Release and ROS Production From the Kupffer Cells.

Considering the possibility that TNF neutralization affects active-oxygen release from Kupffer cells,37 we compared the 2-methyl-6-phenyl-3,7-dihydroimidazo[31]pyrazin-3-one dependent chemiluminescence in plastic-adherent liver MNCs from normal mice and from Con-A–treated mice with or without anti-TNF Ab pretreatment. The chemiluminescent light strength of the adherent Kupffer cells at 1 hour after Con-A administration was stronger than that of the Kupffer cells from normal mice, but light strength declined significantly in the Con-A–administered mice with anti-TNF Ab pretreatment (Fig. 6A).

Figure 6.

The effect of anti-TNF Ab on superoxide (A) or ROS (B) production from Kupffer cells. Data for superoxide production are mean ± SE from four mice in each group. *P < 0.05. Data for intracellular ROS are representative of five independent experiments, which all had similar results.

To further confirm superoxide/ROS production from the CD68+ Kupffer cells, the cells were stained with ROS-sensitive dye, aminophenyl fluorescein, which has same luminous wave length as fluorescein isothiocyanate. In Con-A–administered mice, the ROS content in Kupffer cells increased 1 hour after administration, whereas in mice after anti-TNF Ab pretreatment such an increase of ROS was not observed (Fig. 6B).

The Attenuation of Con-A Hepatitis by Free Radical Scavengers Is Not Accompanied by Cytokine Reduction.

To determine whether oxygen radicals are essential effectors of Con-A hepatitis, we evaluated the effect of a clinically applied free-radical scavenger, Edaravone. Edaravone pretreatment greatly decreased ALT elevation (80% inhibition) after Con-A administration, but cytokine concentrations were hardly affected (Fig. 7A-C). Mice pretreated with GdCl3 showed no change in the capacity for producing cytokines (Fig. 7A–C). Moreover, anti-FasL Ab pretreatment could not reduce the ALT increase (Fig. 7A). Another free radical scavenger, lec-SOD, also decreased ALT levels but did not affect cytokine levels after Con-A injection (Fig. 7D-F).

Figure 7.

Inhibition of hepatic injury (but not serum cytokine levels) by free radical scavengers after Con-A injection. (A, B, C) Mice were pretreated with edaravone, and at the indicated hours after Con-A injection, blood samples were obtained from the retroorbital sinus, and serum ALT and cytokine levels were measured. (D, E, F) Mice were pretreated with edaravone or lec-SOD, and serum ALT and cytokine levels were examined after Con-A injection. ALT, TNF, and IFN-γ levels were examined at 24 hours, 1 hour, and 12 hours, respectively, after Con-A injection. Each group includes at least four mice.

Proportional Change of Kupffer Cells in Anti-NK1.1 Ab–Pretreated Mice.

As reported previously, Con-A hepatitis can be attenuated by the depletion of NKT cells (80% inhibition, Fig. 1), and the involvement of the NKT cells is indispensable for the pathogenesis of this hepatitis.7, 8 Depletion of both NKT and NK cells by the anti-NK1.1 Ab but not depletion of NK cells alone (by anti-asialo-GM1 Ab) partially decreased the increase of CD68+ Kupffer cells after Con-A administration (12.9%) (Fig. 8). However, depletion of both NKT and NK cells did not reduce serum TNF levels at 1 hour after Con-A injection (data not shown).

Figure 8.

Inhibition of the increase in CD68+ Kupffer cells by depletion of NKT cells after Con-A injection. Liver MNCs were obtained 1 hour after Con-A injection from normal mice and mice pretreated with anti-NK1.1 Ab or anti-asialoGM1 Ab. Data are representative of five independent experiments, which all had similar results.

GdCl3 Pretreatment Does Not Decrease TNF Production From Liver MNCs, and CD11b+ Kupffer Cells Produce More TNF Than CD68+ Kupffer Cells After Con-A Stimulation.

Liver MNCs from GdCl3-pretreated mice produced TNF amounts comparable to those of control mice after Con-A stimulation in vitro (Fig. 9A). When F4/80+ cells and F4/80 cells from mice were isolated from liver MNCs by magnetic sorting, and either type of cell (2.5 − 106/mL) was stimulated in vitro with Con-A (5 μg/mL) for 6 hours (TNF levels peaked at 6 or 12 hours after starting culture), the TNF level produced by F4/80+ cells was higher than that of F4/80 cells. However, spleen F4/80+ cells could not produce TNF (Fig. 9A). Because TNF levels produced by Kupffer cells in vitro were relatively low, we also performed ex vivo experiments. Liver MNCs were obtained from mice at 1 hour after Con-A injection, F4/80+ cells and F4/80 cells were isolated and cultured for 6 hours, and the TNF levels in supernatants were examined. It was confirmed that F4/80+ cells are the main TNF producers (Fig. 9B). Finally, CD68+ Kupffer cells and CD11b+ Kupffer cells were examined for their TNF production. Because F4/80+ Kupffer cells were either CD68+ or CD11b+ but NK cells also express CD11b, mice were pretreated with antiasialo-GM1 Ab 3 days before sacrifice to deplete NK cells, and either CD68+ cells or CD11b+ cells were purified from liver MNCs by magnetic sorting (>85% purity) 1 hour after Con-A injection, and were cultured for 6 hours. The results showed that CD11b+ Kupffer cells produced more TNF than CD68+ Kupffer cells (Fig. 9B).

Figure 9.

(A) In vitro TNF production from liver MNCs, liver or spleen F4/80+ cells, and F4/80 cells. Liver MNCs (2.5 × 106) from control mice (L whole) and GdCl3-pretreated mice (Gd whole) were cultured with Con-A (10 μg/mL) in vitro and TNF levels of culture supernatants at 6 hours after starting culture were measured. Liver and spleen MNCs were obtained, and F4/80+ cells and F4/80 cells were cultured and TNF levels of culture supernatants at 6 hours after starting culture were measured. All data are mean ± SE from three independent experiments. (B) Ex vivo TNF production from either liver F4/80+ cells, F4/80 cells, CD68+ Kupffer cells, or CD11b+ Kupffer cells. Mice were pretreated with anti-asialo-GM1 Ab 3 days before sacrifice to deplete NK cells, and either CD68+ cells or CD11b+ cells were purified from liver MNCs by magnetic sorting (>85% purity) 1 hour after Con-A injection. Either cells were cultured for 6 hours, and TNF levels were examined. All data are mean ± SE from three independent experiments.

The Role of Either Con-A–Activated F4/80+ Cells or F4/80 Cells in Hepatic Injury of Normal Mice and NKT-Deficient CD1d−/− Mice.

Control BALB/c mice showed hepatic injury at 12 hours after Con-A injection (milder than B6 mice), and as expected, CD1d−/− mice with BALB/c background did not show hepatitis (Fig. 10A). To further examine which liver MNCs are responsible for Con-A hepatitis, adoptive transfer experiments were performed. Liver MNCs were obtained from normal mice at 1 hour after Con-A injection and were separated into F4/80+ Kupffer cells (5 × 106) or F4/80 lymphocytes (5 × 106) by magnetic sorting and were transferred into mice via spleens, because intravenous injections of either cells could not induce stable liver injury. Unexpectedly, F4/80 lymphocytes (but not F4/80+ Kupffer cells) induced a significant hepatic injury in CD1d−/− mice, whereas F4/80+ Kupffer cells (but not F4/80 lymphocytes) induced a significant hepatic injury in BALB/c mice (Fig. 10B, C). Interestingly, however, GdCl3 pretreatment of CD1d−/− mice abolished the hepatic injury induced by the transfer of F4/80 lymphocytes (Fig. 10D). Furthermore, transfer of F4/80 lymphocytes into non-lethally (4 Gy) irradiated CD1d−/− mice, in which the function of resident Kupffer cells should be suppressed, could not induce hepatic injury (Fig. 10D).

Figure 10.

The effect of adoptive transfer of either F4/80+ cells or F4/80 cells of liver MNCs from Con-A–injected mice into naïve CD1d −/− mice or BALB/c mice. (A) Con-A–induced hepatic injury in BALB/c mice and CD1d−/− mice. At 1 hour after Con-A injection into BALB/c mice, liver MNCs were obtained and either F4/80+ cells or F4/80 cells were isolated by magnetic sorting. (B, C) Cells (5 × 106 of either type) were injected from the spleen into CD1d−/− mice (B) or BALB/c mouse (C), and serum ALT levels were measured at the indicated time points. (D) CD1d−/− mice were treated with GdCl3 (Gd) or irradiation (RAD), and subsequently F4/80 cells (5 × 106) were injected. The data for mice without GdCl3 or irradiation is the same as in B. Data are means ± SE from three independent experiments.

Discussion

The points of the current study are illustrated in a diagram (Fig. 11).

Figure 11.

The interactions between NKT cells, Kupffer cells, TNF, and superoxide in the pathogenesis of Con-A hepatitis.

Activated Kupffer cells produce and release superoxide, which is known to be the main effector of hepatotoxicity induced by either lipopolysaccharide injection37, 38 or by ischemia reperfusion injury.39, 40 Similar to these hepatic injury models, we demonstrated that the increase of CD68+ Kupffer cells and their superoxide/ROS release are the main mechanism of Con-A hepatitis (Fig. 11). Furthermore, we also explored whether GdCl3 abolished hepatic injury by inhibiting the increase of CD68+ Kupffer cells and their superoxide/ROS release. However, it is unclear at present how CD68+ Kupffer cells rapidly increase in the liver after Con-A injection. The proliferation of CD68+ Kupffer cells in situ within 1 hour is highly unlikely. In addition, because spleen macrophages are resistant to GdCl3 treatment,41 and splenectomy reportedly does not inhibit Con-A hepatitis,14 the spleen is unlikely a major source of CD68+ Kupffer cells. Therefore, we speculate that CD68+ Kupffer cells come from CD68 F4/80+ Kupffer cells/precursors in the liver, or they may come from bone marrow. Further study is required to clarify this issue.

TNF did not decrease at all in GdCl3-pretreated mice after Con-A injection, despite the fact that anti-TNF Ab indeed completely inhibited Con-A hepatitis. Consistent with our data, it was previously reported that serum TNF levels and TNF messenger RNA in the liver tissues of Con-A–injected mice were not suppressed by GdCl3 pretreatment.11, 13 It was also reported that although diminution of Kupffer cells by another reagent, clodronate liposomes, did not decrease serum TNF levels, the hepatic injury was abrogated.14 It is speculated that CD11b+ Kupffer cells may compensate for TNF production from CD68+ Kupffer cells in GdCl3-pretreated mice. It is also possible that macrophages of other organs produce TNF. However, spleen macrophages did not produce TNF in vitro after Con-A stimulation, and splenectomy reportedly does not inhibit Con-A hepatitis.14 Conversely, previous reports have suggested that Kupffer cells in rats can be classified morphologically into “large” and “small” subsets41, 42 and that “large” Kupffer cells may selectively disappear on GdCl3 treatment. The TNF-producing cells may be the “small” Kupffer cells in rats. However, in mice, the size of F4/80 cells after Con-A injection did not significantly differ between control and GdCl3-pretreated mice as revealed by forward scatter analysis in flow cytometry, and the CD68+ Kupffer cells and CD11b+ Kupffer cells did not differ in their size either (our unpublished observation).

There has been an additional puzzling phenomenon regarding TNF in Con-A hepatitis. Although anti-TNF Ab pretreatment of mice completely blocks Con-A hepatitis, injection of a sufficient amount of exogenous TNF did not abolish the protective effect of Kupffer cell depletion in mice pretreated with clodronate liposomes.14 Our current results can explain this phenomenon. Namely, even in the presence of TNF, hepatic injury is abrogated if superoxide/ROS-producing Kupffer cells do not increase. Although TNF is required for activation of superoxide/ROS-producing Kupffer cells, neither TNF, activated liver NKT cells, FasL, nor a combination of them can induce Con-A hepatitis. Superoxide/ROS produced by CD68+ Kupffer cells (which are supported by NKT cells and TNF) are thus the essential effector in Con-A hepatitis.

It is well known that NKT cells play an important role in Con-A hepatitis, and they are considered to be a candidate for final effectors.8, 9 Current study demonstrated that the increase of CD68+ Kupffer cells having superoxide-producing capacity was suppressed by the absence of NKT cells. Thus, NKT cells are crucial for the increase of superoxide-producing Kupffer cells. The elucidation of molecules or factors from NKT cells that induce the activation and increase of CD68+ Kupffer cells will be the focus of a future study.

The role of Fas/FasL in Con-A hepatitis has been controversial. However, the current results demonstrated that although anti-FasL Ab suppressed hepatic injury induced by α-GalCer, it did not inhibit Con-A hepatitis, supporting the idea that Fas/FasL is not essential in Con-A hepatitis. A typical FasL-mediated hepatic injury is observed when the NKT cell ligand α-GalCer is injected into aged mice, in which FasL of ligand-activated NKT cells induced by TNF (from Kupffer cells) is directly responsible for hepatic injury.18, 32, 33, 43 The current results in α-GalCer hepatitis also confirmed this TNF/FasL/Fas system. Furthermore, we also demonstrated that α-GalCer hepatitis was not inhibited at all by GdCl3 pretreatment, presumably because GdCl3 pretreatment did not abolish TNF production.

Of particular interest, adoptive transfer of Con-A–activated F4/80 lymphocytes (but not F4/80+ Kupffer cells) induced hepatic injury in CD1d−/− mice, whereas F4/80+ cells (but not F4/80 lymphocytes) induced hepatic injury in normal mice. Furthermore, removal or inactivation of CD68+ Kupffer cells in CD1d−/− mice abolished hepatic injury induced by transferred-F4/80 lymphocytes, supporting the idea that activated NKT cells (and T cells) alone cannot induce hepatocyte injury without intact Kupffer cells. However, in a similar sense, activated Kupffer cells alone cannot induce hepatic injury in CD1d−/− mice either. It is speculated that transferred NKT cells (and CD4+ T cells) may be suppressed in their activation by resident liver leukocytes in normal mice, but transferred Kupffer cells can induce the activation of resident NKT cells (and CD4+ T cells) and cooperate in causing hepatocyte injury. In contrast, NKT-deficient CD1d−/− mice accept the activated NKT cells and they may further stimulate resident Kupffer cells in a positive feedback loop and induce hepatocyte injury. However, Kupffer cells transferred into CD1d−/− mice could not lead the further activation of themselves and could not induce hepatocyte injury because of the absence of NKT cells.

Taken together, we conclude that Con-A hepatitis is a superoxide/ROS-dependent hepatic injury and that a close interaction of Kupffer cells with NKT cells is important for Con-A hepatitis.

Ancillary