A murine model of acute liver injury induced by human monoclonal autoantibody


  • Potential conflict of interest: Nothing to report.


We have previously reported an immunoglobulin (Ig) M autoantibody to hepatocyte-related 190-kd molecules in patients with type 1 autoimmune hepatitis (AIH). This molecule was first isolated by hepatocyte-specific human monoclonal antibody (MoAb). To elucidate the role of this IgM autoantibody in hepatocyte injury, we examined the reactivity of this MoAb to murine hepatocytes and then questioned whether acute hepatic injury could be induced in mice via injection of this MoAb. The reactivity of MoAb was examined via both FACS analysis using murine hepatocytes and immunostaining of liver tissues. We then identified the murine hepatocyte membrane molecule recognized by this MoAb. The role of this MoAb in the immunopathogenesis of AIH was assessed by testing whether its injection into mice could increase serum aminotransferase levels as well as cause changes in liver histology. The present results demonstrate that this MoAb cross-reacted with murine hepatocytes and recognized a 190-kd molecule on the murine hepatocyte membrane just as in human hepatocytes. One hour after the injection of MoAb, the deposition of both IgM and complement component 3 was found in liver tissues. At 8 hours after the injection, serum aminotransferase levels were significantly increased in MoAb-injected mice compared with controls. Histological study revealed massive hepatocyte necrosis in MoAb-injected mice. In conclusion, human MoAb recognized a 190-kd molecule of both human and murine hepatocytes, and the injection of this MoAb to mice resulted in acute liver injury, indicating that this type of autoantibody may play an important role in the immunopathogenesis of AIH. (HEPATOLOGY 2005.)

Autoimmune hepatitis (AIH) is a chronic disorder of hepatocytes that is accompanied by immunological abnormality as determined by serological examination.1, 2 Based on the observation of alterations in several T cell functions, AIH is generally thought to be a disorder mediated by T cells.3–6 However, various types of hepatocyte-specific autoantibodies frequently found in AIH patients are also suspected of being involved in immune-mediated liver injury in this disease. This was supported by the following results: (1) Serum autoantibodies obtained from AIH patients were able to kill human hepatocytes,7–12 (2) the titer of hepatocyte-specific autoantibody was tightly correlated with liver damage in AIH patients.13, 14 However, despite these important observations, because of the lack of an adequate in vivo system, the precise role of such autoantibodies in immunomediated liver injury has remained obscure.

We have previously reported the hepatocyte-specific human monoclonal autoantibody (MoAb) produced by human clone established by peripheral blood mononuclear cells of patients with AIH.15 This MoAb recognized a 190-kd molecule on the hepatocyte membrane and was able to kill human hepatocyte cell lines in a complement-dependent manner. In addition, a similar 190-kd molecule-recognizing immunoglobulin M (IgM) autoantibody was also found in AIH patients, and its titer was tightly correlated with serum alanine aminotransferase level. Based on these results, we predicted that this autoantibody might be involved in the immunopathogenesis of AIH. Because human hepatocyte-specific autoantibodies tended to cross-react with rodent hepatocytes,16, 17 we first examined whether this human MoAb would recognize murine hepatocytes. Subsequently, its possible role in the immunopathogenesis of AIH was examined in an in vivo murine model.


Ig, immunoglobulin; AIH, autoimmune hepatitis; MoAb, monoclonal autoantibody; MP, myeloma protein; TBS, Tris-buffered saline; C3, complement component 3; AST, asparate aminotransferase; ALT, alanine aminotransferase.

Materials and Methods

Preparation of Human Monoclonal Autoantibody.

We have established an Epstein-Barr virus–infected human clone using peripheral blood mononuclear cells obtained from a patient with clinical features fulfilling the diagnostic criteria of AIH.18 This clone continuously produces IgM (γ-type)-class hepatocyte-specific MoAb.15 We purified MoAb in a culture supernatant of this clone by using Hitrap IgM purification (Amersham Pharmacia Biotech AB, Uppsala, Sweden) according to the manufacturer's instructions. Contaminating endotoxin in these samples was further removed using Detoxi-Gel (Pierce Co., Rockford, IL). This column showed that the amount of endotoxin was less than 5 pg/mL. Control IgM myeloma protein (MP) was obtained from the serum of a patient with IgM myeloma by the same purification method.

Flow Cytometry Analysis and Immunostaining of Liver Tissues.

Female Balb/c mice, 6 weeks old, were purchased from Nihon Kurea Co. (Tokyo, Japan). Murine hepatocytes were isolated from liver tissues according to the previously described method19 and were used as targets for FACS analysis. In brief, 1 × 105 cells were incubated at 37°C for 60 minutes with either MoAb or MP, washed, and further incubated with 5 μL of 1:10 diluted FITC-conjugated goat anti-human IgM antibody (Biosource Co., Camarillo, CA) for 30 minutes at 4°C. Positively stained cells were then analyzed via FACScan.

Murine liver tissues were embedded in O.C.T. compound (Tissue ek II, Miles Laboratories, Naperville, IL) and frozen in liquid nitrogen. Four-micrometer-thick sections obtained with a cryostat microtome were reacted with either MoAb or MP at 37°C for 120 minutes and then processed for reaction with FITC-labeled anti-human IgM at 37°C for 1 hour. They were observed under a fluorescene microscope.

Immunoblot Analysis.

The molecular features of this MoAb-recognizing murine hepatocyte membrane molecule were examined using a previously described method.15 In brief, for each sample, 50-100 μg of membrane protein was resolved on SDS 10% polyacrylamide gels and transferred to nitrocellulose filters using an electroblotting transfer system. Filters were incubated for 2 hours in Tris-buffered saline (TBS) containing 5% nonfat dry milk at room temperature, and then overnight with primary antibody to specific proteins in TBS with 3% nonfat dry milk containing sodium azide. Filters were washed four times for 5 minutes each in TBS 0.2% Tween 20 and then incubated with secondary antibodies in TBS for 1 hour and washed five times for 10 minutes each in TBS 0.2% Tween 20. Immune complexes were visualized via enhanced chemiluminescence. We examined tissue samples obtained from murine liver and compared them with huH-7 (human hepatocellular carcinoma cell line) lysates.

In Vivo Studies.

To determine whether the injected MoAb can bind to murine liver tissue, we injected 400 μg of either MoAb or MP intraperitoneally; 60 minutes later, these liver tissues were stained with either FITC-labeled anti-human IgM or FITC-labeled anti-human IgG (Biosource). Simultaneously, the same sample liver tissues were also stained with FITC-labeled anti-mouse complement component 3 (C3) (ICN Pharmaceuticals Inc., Aurora, OH). Subsequently, we injected 8 mice with either MoAb or MP (400 μg in 1.2 mL phosphate-buffered saline), and 5 mice were injected only with 1.2 mL phosphate-buffered saline. The serum enzyme activities of both asparate aminotransferase (AST) and alanine aminotransferase (ALT) were serially examined by routine laboratory tests. Differences in serum AST and ALT levels among groups were examined using the Student t test. A P value of less than .05 was considered to be significant in all results. Eventually, 2 of 8 mice in both MoAb- and MP-injected cohorts were killed at 8, 24, and 72 hours, respectively, after the injection to examine liver histology by light microscopy.


Reactivity of Human MoAb With Murine Hepatocytes.

As shown in Fig. 1, isolated murine hepatocytes were positively stained with MoAb (Fig. 1A) but not with MP (Fig. 1B), indicating that our hepatocyte-specific human MoAb can react with murine hepatocytes. These results were further confirmed via immunohistochemistry using normal murine liver tissues. The results shown in Fig. 1C reveal that murine liver tissues reacted to this MoAb (compared with Fig. 1D). The linear staining pattern of the hepatocyte membrane was quite similar to that of human liver tissues.15 These results show that this human MoAb, which is specific for hepatocytes, cross-reacts with murine hepatocytes.

Figure 1.

Reactivity of MoAb with murine hepatocytes. The isolated hepatocytes were stained with MoAb (A) or MP (B) and analyzed using FACS. (C-D) Immune histochemistry of murine liver tissues (original magnification ×25). Note that the cell membrane was positively stained with MoAb (C) but not with MP (D). CV, central vein.

Molecular Features of MoAb-Recognizing Molecule.

This MoAb-recognizing molecule was examined via immunoblotting. As shown in Fig. 2A, the same 190-kd molecule was identified in the samples obtained from both huH-7 (human hepatocellular carcinoma cell line) and murine liver tissues. This molecule was not observed in the same lysates using control MP instead of MoAb (Fig. 2B).

Figure 2.

Immunoblot analysis of MoAb-recognizing molecule. Cell lysates were obtained from huH-7 cells and murine liver tissues. After their protein amounts were adjusted, immunoblotting was performed using MoAb (A) or MP (B). Note that a 190-kd molecule was found in huH-7 and murine liver tissues but not in the same samples using MP. MoAb, monoclonal autoantibody; MP, myeloma protein.

Immunostaining of Liver Tissues Obtained From MoAb-Injected Mouse.

We then examined whether in vivo injected MoAb could directly bind to murine hepatocyte membrane. As shown in Fig. 3, 1 hour after the injection of MoAb, the hepatocyte membrane was positively stained with FITC-labeled anti-human IgM (Fig. 3A) but not with FITC–anti-human IgG (Fig. 3B). Such IgM deposition was not found in other tissues of the same mouse (data not shown) or liver tissues obtained from MP-injected mice (Fig. 3C). Furthermore, we have also found that the same liver tissues obtained from MoAb-injected mice were positively stained with FITC-labeled anti-C3 (Fig. 3D). C3 deposition was not found in the liver tissues obtained from MP-injected mice (Fig. 3E). Thus, both of the injected MoAb and the C3 component were found to have bound to the hepatocyte membrane by 1 hour after the injection.

Figure 3.

Liver tissues from MoAb-injected mice at 60 minutes after the injection were positively stained with FITC-labeled anti-human IgM (A) but not IgG (C). Liver tissues from MP-injected mice were not stained with FITC–anti-human IgM (D). (B,E) C3 deposition was found only in MoAb-injected liver tissues, not in MP-injected murine liver (original magnification ×25). CV, central vein; C3, complement component 3.

Induction of Acute Hepatic Injury in Mouse by MoAb Injection.

Based on the above results, we injected 400 μg of MoAb, MP or phosphate-buffered saline into mice and then examined their serum ALT and AST levels. Figure 4 shows that when mice were injected with MoAb, significant elevation of serum ALT and AST were observed, reaching their maximum level at 8 hours postinjection. However, the elevation in serum ALT and AST levels was much weaker in MP-injected mice. The differences in serum ALT and AST between MoAb-injected mice and MP-injected mice were statistically significant (P < .05 at 4, 8, and 12 hours postinjection).

Figure 4.

Serum AST/ALT levels in MoAb-injected mice. Both AST and ALT levels were significantly elevated in MoAb-injected mice, reaching their maximum level at 8 hours postinjection. The differences in AST/ALT levels between MoAb-injected mice and controls were statistically significant at 4, 8, and 12 hours postinjection. Numbers in parentheses indicate the number of mice.

Histological Features of MoAb-Injected Mice.

During the above study, we also examined the histological features of liver tissues from both MoAb- and MP-injected mice. The results shown in Fig. 5 clearly demonstrate that numerous massive hepatic necrotic lesions were apparent at 8 hours after the injection of MoAb (Fig. 5A). Histological changes of MoAb-injected mice were found in both pericentral (Fig. 5C) and periportal areas (Fig. 5D). Although the histological changes were less prominent, similar results were found in MoAb-injected mice at 24 hours and 72 hours postinjection (data not shown). On the contrary, such necrotic lesions were not found in the livers obtained from MP-injected mice (Fig. 5B).

Figure 5.

Liver histology of MoAb-injected mice. Hepatocytes of the MoAb-injected mouse show prominent eosinophilic degeneration with cell dropout. (A [MoAb-injected liver] vs. B [MP-injected liver]). (C–D) Massive hepatocyte necrosis around the periportal area (C) and pericentral area (D) (original magnification ×25 [A–B], ×50 [C–D]). CV, central vein; PV, portal vein.

Histological examination of other tissues (e.g., stomach, colon, kidney, and lung) obtained at 8 hours postinjection revealed no significant inflammatory changes in MoAb-injected mice (Fig. 6A–D vs. Fig. 6E–H). Therefore, it was clear that the histological injury induced by this MoAb was restricted to the liver.

Figure 6.

Histological features of nonliver tissues of MoAb-injected mice. No inflammatory changes were found in any tissues, including the stomach, colon, kidney, and lung (upper panels show tissues from MoAb-injected mice; lower panels show tissues from MP-injected mice) (original magnification ×25). MoAb, monoclonal autoantibody; MP, myeloma protein.


Regarding the precise mechanisms of immunomediated liver injury in AIH, several possibilities have been proposed. One is that liver-specific autoantibodies might have an important role in causing liver injury in either a complement-dependent manner or via antibody-dependent cytotoxic cells.7–11 Another suggested that T cells play a crucial role in hepatocyte injury in AIH.3–6 In the previous study, we reported the features of liver-specific human MoAb produced from Epstein-Barr virus–infected peripheral blood mononuclear cells obtained from an AIH patient, and hypothesized that this MoAb might be involved in immunomediated liver injury in this disease.15 In the present study, we found that this human MoAb reacted with murine hepatocytes membrane as with that of human hepatocytes (see Fig. 1). In addition, by way of immunblot assay, we demonstrated that this MoAb reacted with a 190-kd molecule of murine hepatocyte lysate just as with that of human hepatocyte lysate (see Fig. 2). All these results have demonstrated that human MoAb recognized a 190-kd molecule that is expressed on the hepatocyte membrane in both humans and mice.

Taking advantage of these features of this MoAb, we then addressed the issues of whether it is directly related to hepatocyte injury. In the present study, we found that 60 minutes after the injection of this MoAb into mice, a deposition of IgM occurred in the liver (see Fig. 3). In addition, the same liver tissue was also stained with FITC-labeled anti-C3. Such phenomena were not observed in the control MP-injected murine livers. These results reveal that by way of intraperitoneal injection, this MoAb specifically binds to hepatocyte membrane together with complement. Then, as shown in Fig. 4, at 4 to 12 hours postinjection, serum ALT and AST exhibited significant elevations. In contrast, injected control MP resulted in only slight elevations in serum ALT and AST. Liver histology of the MoAb-injected mice showed massive hepatocyte injury around the central vein and in the periportal area at 8 hours postinjection (see Fig. 5A,C-D) and a spotty necrosis at 72 hours postinjection (data not shown). In the typical human AIH liver, periportal necrosis and cellular infiltration in portal triads are dominated.20 However, it has been previously reported that pericentral necrosis was occasionally found at a relatively early stage in AIH patients.21, 22 Furthermore, the simple injection of MoAb may account for the absence of cellular infiltration of the portal triads in this model. Based on this assumption, we can suppose that the histological changes found in MoAb-injected mice might at least partially resemble those of human AIH livers. This MoAb-induced histological damage was only found in liver tissues, not in nonliver tissues such as the stomach, colon, kidney, and lung (see Fig. 6). The same results of elevated serum AST and ALT levels and massive hepatocyte necrosis were also found in other MoAb-injected mice, such as C57Bl/6 (data not shown).

At least in this murine model, antibody-dependent cytotoxic cells apparently did not contribute to cell damage, because we did not observe cellular infiltration as a prominent histological feature. Rather, the deposition of C3 in the same liver tissues suggests that complement-mediated cell cytotoxicity might in fact be the responsible mechanism in this in vivo model. As has been previously reported,23, 24 many Caucasian AIH patients have deletions in C4A and C5B loci, resulting in a reduced C4 level in relatively younger AIH patients. A C4A loci deletion is known to be in strong linkage disequilibrium with HLA-A1, B8, DR3, DR52a, and DQ2, of all of which are associated with susceptibility to AIH in Caucasian AIH patients. However, many Japanese AIH patients showed strong linkage to HLA-DR4 but not to HLA-DR3.25 This result will explain why such C4A deletion was not seen in many Japanese AIH patients (personal communications from Drs. M. Zeniya and G. Tada, Tikei University School of Medicine, September 2004). This finding indicated that genetic backgrounds, including low C4 concentration, are not identical between Japanese and Caucasian patients. Rather, it was previously reported that complement components were successfully detected in the livers of several AIH patients.9 These results may support our prediction that a complement-dependent cytotoxicity may have some role in the immunopathogenesis of this disease.

A similar MoAb-mediated liver cell injury, using rat hepatocyte-specific murine MoAb, was reported by Ikeda and Kurebayashi.26 Although their MoAb recognized a 98-kd molecule (rat liver-specific antigen [RLSA]),27 the same complement-mediated cell lysis may have contributed to hepatocyte injury in the MoAb-injected rat. The liver histology observed in the MoAb-injected mouse was quite similar to that in the anti-Fas–injected mouse.28 However, the molecular features recognized by this MoAb were not identical to those of the Fas antigen, and this MoAb reacted with the Fas-negative cell line huH-7. In addition, several inflammatory cytokines, such as interleukin 1a, interleukin 1b, tumor necrosis factor α, and interleukin 6, were hardly detected in the purified injected MoAb (data not shown). Thus it can be concluded that the acute liver injury induced by this MoAb was not due to such nonspecific cytokines. Rather, all the results presented in this study clearly demonstrated that human autoantibody directed to the 190-kd molecule on the hepatocyte membrane is able to induce hepatocyte injury in vivo. It has been reported that autoantibody-mediated cytotoxicity is important in tissue damage in certain organ-specific autoimmune diseases.29, 30 As was reported in our previous study, a similar 190-kd molecule-recognizing IgM autoantibody was frequently found in AIH patients, and its titer was tightly associated with serum ALT levels.15 The present study, as well as these previous results, suggest that the 190-kd molecule on the hepatocyte membrane might be one of the target molecules for the autoimmunity that is responsible for liver injury in AIH. The more precise features of this molecule (e.g., the amino acid sequence) are still unclear, though their clarification is currently underway in our facilities.

We are still unable to eliminate the possibility that other mechanisms, such as T cell–dependent liver injury, may also have an important role in the immunopathogenesis of AIH. As was predicted by Nagaraju et al., both autoantibody-dependent and T cell–dependent mechanisms may be centrally implicated in organ injury in MS.31 Therefore, the same phenomenon might be involved in immunomediated liver injury in AIH. This possibility will have to be carefully examined.


The authors thank Dr. T. Ezaki, Department of Anatomy and Developmental Biology, Tokyo Women's Medical University, for his thoughtful help, and A. Takei for her excellent secretarial assistance.