These authors contributed equally to the study.
Agonistic antibodies to Fas induce a breach in the endothelial lining of the liver and a breakdown in B cell tolerance
Article first published online: 21 DEC 2006
Clinical & Experimental Immunology
Volume 147, Issue 2, pages 346–351, February 2007
How to Cite
Newkirk, M. M., Nowak, U., Skamene, E., Iera, D. and Desbarats, J. (2007), Agonistic antibodies to Fas induce a breach in the endothelial lining of the liver and a breakdown in B cell tolerance. Clinical & Experimental Immunology, 147: 346–351. doi: 10.1111/j.1365-2249.2006.03279.x
- Issue published online: 5 JAN 2007
- Article first published online: 21 DEC 2006
- Accepted for publication 9 November 2006
- liver damage;
- rheumatoid factor
Liver disease can be associated with a breakdown in self-tolerance and the production of autoantibodies such as rheumatoid factors (RF), which bind to IgG. Here we investigated whether primary, non-infectious liver damage was sufficient to induce autoantibody production. We established a model of targeted liver damage induced by weekly sublethal injections of pro-apoptotic anti-Fas (CD95) antibodies. Liver damage, monitored by measurements of serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels, was minimal 1 week after anti-Fas injection. However, the sublethal Fas stimulation was sufficient to trigger significant haemorrhage in the liver, as assessed by Evans Blue dye leakage into the organ 5 h after anti-Fas antibody injection. We observed an induction of RF in response to the weekly injections of sublethal anti-Fas antibodies but not of isotype control antibodies, indicating a breakdown of self-tolerance induced by Fas engagement. RF induction was unlikely to be due to direct activation of B cells, as splenocytes stimulated with anti-Fas antibodies in vitro did not produce RF. These studies show that sublethal damage to the liver by Fas engagement leads to liver haemorrhage and is sufficient to trigger the breakdown of self-tolerance.
The liver is a unique organ that contains lymphocytes, B cells and T cells, but does not normally harbour a constitutive inflammatory response. Although capable of mounting an immune response to pathogens such as the hepatitis viruses, the liver favours tolerance over immunity  and is considered immune privileged . The sinusoidal endothelial cells (SECs) play a key role in the maintenance of privilege. The liver also has an extraordinary ability to regenerate in response to toxins, pathogens or surgery. Mice that undergo partial hepatectomy of > 50% can regain their entire liver mass within 7 days . Hepatic apoptosis has been shown to occur in a number of liver diseases including alcoholic liver injury, acute liver failure and even with hepatitis C viral (HCV) infections [4–6]. However, with the latter, the virus has evolved a strategy to block one of the pathways of apoptosis, namely that via tumour necrosis factor (TNF)-α[7,8], but leaving the Fas pathway intact.
As a model system for liver damage, much information has been derived from targeting Fas with selective antibodies. Depending on the specific antibodies used, the dose and the mouse strain, such treatment can be lethal, with death occurring 6–12 h after injection . The liver appears to be the primary organ targeted, and the cause of death is fulminant hepatic failure . Although both SECs and hepatocytes can express Fas, it appears that the former cells are critical to the lethality caused by antibodies such as Jo2, although hepatocytes are also killed by agonistic anti-Fas antibodies . Jodo  and Xu  have shown Jo2 anti-Fas antibody targets the SECs that express FcγRII. The FcR appears to anchor the anti-Fas antibodies in the sinusoidal cells. The net outcome of this anti-Fas antibody binding is haemorrhaging, which leads subsequently to fulminant liver failure and contributes to inducing hepatocyte apoptosis.
In hepatitis C virus (HCV) infections tolerance can be broken, resulting in autoantibody production. An autoantibody detected frequently in the serum of HCV patients is rheumatoid factor (RF) [13,14], which binds IgG in the γ2–γ3 cleft . RF can be a part of the normal immune response to a variety of pathogens . RF are an essential component of the mixed cryoglobulin (MC) that is detected in 30–40% of patients with HCV. MC appear late in HCV infection, and are associated with more extensive liver damage. MC contribute to the major extra-hepatic manifestations of HCV [16,17].
We hypothesized that liver damage could directly induce the breakdown of self-tolerance and appearance of RF, even in the absence of HCV-mediated immune stimulation. Here we report that the breakdown of B cell tolerance follows anti-Fas antibody-induced liver damage. When mice were injected with sublethal amounts of Jo2 anti-Fas antibodies we observed liver haemorrhage and induction of RF, which correlated directly with liver damage. As anti-Fas antibodies were not found to activate B cells directly in vitro, we speculate that IgG entering the liver via Fas-mediated breaches in the endothelial cell barrier probably activated the resident RF-producing B cells by stimulation of the B cell receptor.
Materials and methods
C57Bl/6 mice (Charles River, Montreal, Canada) were maintained in a conventional facility and used at 4–6 weeks old for all experiments except the in vitro splenocytes activation, where C3H/HeJ mice (Jackson Laboratory, Bar Harbour, Maine, USA) were used. All studies were approved by the McGill University Animal Care Committee.
Injection of mice with anti-Fas antibodies and Escherichia coli glycolipoprotein (GLP)
Groups of mice (n = 4–6) were injected intraperitoneally (i.p.) at weekly intervals for 4 weeks with 200 µl sterile phosphate-buffered saline (PBS), containing 4 µg/mouse isotype control or anti-Fas antibody (half lethal dose, clone Jo2, BD Biosciences, Mississauga, ON, Canada) and/or 50 µg/mouse Escherichia coli (E. coli) glycolipoprotein (GLP). GLP, used as a positive control for RF induction, was semipurified from 24-h cultures of E. coli propagated in LB broth as a stagnant culture by precipitation in 40% ammonium sulphate and gel filtration in an S300 column and screened for RF induction in vitro. Peak 1 consistently induced RF and was used in the experiments. Purity of GLP as determined by silver staining revealed two to three bands at ∼300–400 kDa, which migrate very close to one another. Carbohydrate staining indicated a broad band that encompassed the three bands. One group of mice was untreated. Previous studies in the laboratory had shown that GLP induced RF both in vivo and in vitro and was thus included as a positive control. Mice receiving both anti-Fas antibodies or isotype control antibody and E. coli GLP did not receive GLP in week 2. Mice were weighed and bled weekly from the saphenous vein and blood was collected by cardiac puncture at euthanasia.
Organ haemorrhage as measured by Evans Blue dye (EBD) leakage
Mice (seven per group) were injected with Jo2 anti-Fas antibodies or isotype control as above, and 5 h later mice were injected in the tail vein with 3 µl/g EBD. The mice were anaesthetized after 10 min and then perfused for 10 min with PBS. Tissues were harvested, dissected, weighed and divided in two equal sections. The first portion was desiccated at 60°C for 24 h and weighed (dry tissue) and the second portion was extracted in formamide (4 ml/g wet tissue weight) at 24°C for 24 h. After centrifugation of the extracted portion (3000 g, 20 min), the absorbance of EBD was measured by spectrophotometry at 620 nm. EBD was expressed as OD620 EBD per g dry tissue.
Detection of RF
IgM RF were detected as described previously . All samples were tested in duplicate in two independent experiments with appropriate positive and negative controls.
Measurement of total IgM
Total IgM was measured by enzyme-linked immunosorbent assay (ELISA) as described previously , with slight modifications. ELISA plates were coated with 100 µl/well of a 3-µg/ml goat anti-mouse IgM + IgG (Jackson ImmunoResearch, West Grove, PA, USA) in 0·05 M carbonate–bicarbonate (pH 9·6) overnight at 4°C. Bound IgM was detected using 100 µl/well horseradish peroxidase (HRP)-conjugated goat anti-mouse IgM (Jackson ImmunoResearch) diluted 1/8000 in PBS-Tween. The fact that the anti-sera captures both IgM and IgG on the plate does not compromise in any way the ability to measure total IgM, as the detecting antibody is specific for IgM only. A standard curve was generated using mouse IgM (BD Biosciences). All samples were tested in duplicate in two independent experiments with appropriate positive and negative controls.
Blood urea nitrogen (BUN)
BUN was measured by a colorimetric assay (ThermoTrace, Trace Scientific Ltd, Melbourne, Australia) following the manufacturer's method. Assays were performed in duplicate and the concentration was calculated based on a known standard (Sigma-Aldrich, St. Louis, MO, USA).
Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels were analysed by a multi-channel autoanalyser in the clinical chemistry department of the Montreal General Hospital. Only the blood collected by cardiac puncture was analysed due to the requirement for large volumes of serum. It was not possible to measure these enzymes in the EBD study, as perfusion depleted the mice of serum.
Spleens were collected for cell preparations from C3H/HeJ mice. The spleens were made into single-cell suspensions by the gentle grinding of tissue between frosted glass slides and the RBC were lysed with 0·83% ammonium chloride. The cells were cultured for 3 days in Dulbecco's modified Eagle's medium (DMEM) containing 10% Nu-serum replacement (BD Biosciences) at a density of 1 × 106 cells/ml in round-bottomed culture tubes in a CO2 incubator. For the in vitro cultures, Jo2 anti-Fas or isotype control was added at a concentration of 2 µg/ml with or without E. coli GLP (10 µg/ml). Control tubes contained media alone (negative control) or media plus GLP (positive control). Cultures were performed in duplicate multiple times. At the end of the culture period the supernatant was harvested for RF measurement and the number of viable and dead cells were counted using trypan blue.
Data were analysed using InStat2 (GraphPad, San Diego, CA, USA). Student's t-test or the Mann–Whitney U-test was used as appropriate.
Weight gains and liver enzyme levels did not differ significantly between groups of mice injected with anti-Fas or isotype control antibodies (data not shown). This probably reflects the efficient regenerative capacity of the liver, as the liver enzymes were analysed at euthanasia, 1 week after the final antibody injection. In other studies, mice that received the same dose of anti-Fas had only minimally elevated levels of liver enzymes 8 h after injection .
To determine whether the sublethal amounts of anti-Fas antibody induced liver haemorrhage, EBD leakage was measured as a marker of the damage to the liver endothelial cell barrier. In mice injected with anti-Fas antibody, but not in isotype control mice, there was a significant leakage of EBD, injected 5 h after antibody administration and 10 min prior to euthanasia, detected in the liver (P = 0·007) after removal of all venous blood (Fig. 1). There appeared to be variation in the anti-Fas treated mice, which might reflect a natural variation in responsiveness to the anti-Fas or the injection technique. All the tail vein injections were performed by the same individual to limit variation. No significant difference in the amount of EBD was detected in heart, lungs or spleen 5 h after injection of anti-Fas antibody (data not shown), whereas in the kidney a small but significant increase in EBD was detected after anti-Fas antibody injection, but not with the isotype control (P = 0·05).
Knowing that the liver harbours RF-producing B cells , we determined whether sublethal levels of anti-Fas antibodies would induce RF. Weekly injection of agonistic anti-Fas antibody for 4 weeks significantly induced IgM RF by week 3, compared to the prebleed levels (P = 0·033) (Fig. 2a), and was even more significant at week 4 (P = 0·0033). At week 4 there was a correlation between RF induction and AST (r = 0·601) and ALT (r = 0·888) (Fig. 2c). The RF increase was not observed in untreated or isotype control antibody-treated mice, thus indicating that RF induction was not due to non-specific systemic B cell activation via Fc receptors (FcR). Total IgM increased in anti-Fas-injected mice compared to control mice at week 3 (P = 0·0001) (Fig. 2b). However, the fold increase was 1·3 compared to an 8·5-fold increase in IgM RF, demonstrating antigen-specific activation of RF-producing B cells and not simply an increase due to polyclonal activation of all IgM+ B cells. The kinetics of the RF response after anti-Fas antibody injection suggested stimulation of naive B cells.
As a positive control for RF induction, a GLP macromolecule from E. coli that signals through the Toll-like receptor pathway and requires MyD88 (data not shown) was contrasted to anti-Fas antibody treatment. GLP induced RF by 5 days after the first injection. This was not due to FcR engagement by the control antibody injected along with GLP, as there was no significant increase in IgM RF for mice injected with control antibody alone (Fig. 2d). This rapid induction of RF suggests a memory response, which was not surprising as most mice harbour E. coli in the gut. GLP caused only a transient increase in total serum IgM, which was not significant at week 4 (Fig. 2e).
To determine whether anti-Fas antibody stimulated RF production by binding to RF-producing B cells directly, primary splenocytes from C3H/HeJ mice were cultured with anti-Fas antibodies in vitro. We have found previously that splenocytes from C56Bl/6 mice produced only verynegligible amounts of RF in vitro, even when stimulated with a potent RF inducer such as GLP isolated from Proteus mirabilis. In contrast, we had found that C3H/HeJ mice produced the most RF both in vivo and in vitro, consistent with previous publications . Culturing C3H/HeJ splenocytes for 5 days with anti-Fas antibodies did not stimulate RF production when compared to untreated cells or cells treated with control antibody (Fig. 3). In contrast, incubation with E. coli GLP induced significant levels of RF. When cells were incubated with GLP plus anti-Fas antibodies, the RF response decreased, but this difference did not reach statistical significance. There was no difference in RF production by splenocytes cultured with GLP plus control antibody, confirming that RF secretion was not due to FcR engagement. There appeared to be a slight increase in B cell apoptosis induced by anti-Fas antibody, as the percentage of trypan blue-positive cells after 2 days of culture was 83 ± 9 (n = 9) for the splenocytes incubated with anti-Fas antibody, compared to 75 ± 9 (n = 9) for the isotype control cultures.
Liver injury induced through the Fas pathway has been relatively well defined [9,10,22,23]. However, little has been conducted to investigate the long-term consequences of sublethal Fas-induced damage. We show that sublethal exposure to a Fas agonist caused liver injury and that this can lead to the loss of self-tolerance and the induction of RF.
Anti-Fas antibodies bind sinusoidal endothelial cells, causing liver haemorrhage and lethality in mice [11,12]. Our study showed that weekly injection of half the lethal dose of Jo2 also caused a breach in the endothelial lining of the liver, sufficient to allow EBD to leak into the tissue. Other studies have investigated the impact of varying amounts of Jo2 antibody, and have shown that there was only a minimal increase in ALT and AST and minor changes in liver morphology with the dose used here when measured 8 h after injection of the anti-Fas antibody . However, when combined with an additional insult (in their case pyrazole) the damage to the liver increased to significant levels due to the apoptosis induced by the sublethal levels of anti-Fas antibodies. In our studies, no abnormalities in liver enzymes were detected at euthanasia in mice injected with the sublethal anti-Fas levels. This may, in part, reflect the efficient rate of liver regeneration. Previous studies have shown that liver enzymes are markedly increased when measured from 2 to 7 h post-injection of a lethal dose of Jo2 [10,24]. Due to the requirement of large serum volumes, it was possible to determine liver enzymes levels in our study only at the terminal bleed.
From our studies using the EBD methodology it appears that haemorrhage in the kidney also occurred subsequent to the injection of anti-Fas antibody, but no leakage was detected in the spleen. In imaging studies, Zinn  has shown that the 99mTc-Jo2 anti-Fas antibody at 4 µg/mouse is taken up rapidly by the endothelial cells in the liver with one pass through the circulatory system (by 0·2 h after intravenous injection the liver endothelial cells were saturated, with slightly slower kinetics for intraperitoneal injections); however, they also detected some uptake in the spleen of BALB/c mice. In our case, it appears that if the anti-Fas antibodies did bind to cells in the spleen no major damage occurred, such that EBD leakage could be detected. The EBD methodology has the advantage of determining the consequences downstream of the apoptotic event induced by the Jo2 anti-Fas antibodies in vivo.
Importantly, our studies suggested that breaching the endothelial lining of the liver may contribute to a loss of self-tolerance and the production of IgM RF autoantibodies. The response became measurable only by week 3, consistent with activation of naive B cells. The induction of RF by anti-Fas antibodies was probably not due to direct activation of RF-producing B cells, as culturing splenocytes in vitro with anti-Fas did not stimulate RF production. We also show that the activation was not due to the interaction of the anti-Fas antibodies with Fc receptors, as no RF production was detected in mice injected with isotype control antibodies.
The precise mechanism for the stimulation of the RF response is as yet unknown, but it is likely that because B cells do reside in the liver [20,26,27] the B cell receptor of the RF-producing cells could be cross-linked by aggregated IgG, which would lead to the production of RF, if IgG aggregates penetrate the liver parenchyma following damage to the endothelium. Monomeric IgG is not efficient at triggering the RF response, whereas immune complexes are . It is possible that there is some RF produced from extrahepatic B cells in the anti-Fas-treated mice, but the major source is likely to be the liver, as that is the site of damage and there is a correlation between the liver enzymes AST and ALT and RF production. The number of RF-producing B cells in the liver is unknown, but is likely to be between the percentage found in circulation, roughly 1% of B cells , and that in normal spleen, 10–15% of B cells [29,30]. These represent high percentages for one specificity of B cells, suggesting that RF production could be induced relatively readily with the loss of self-tolerance.
In contrast to RF induction by anti-Fas antibodies, treatment with E. coli GLP caused a rapid induction of RF by 5 days after injection, indicative of an IgM memory response. This is not unexpected, as E. coli is resident in the gut flora and GLP complexed with IgG may activate RF B cells. TLR ligation probably contributes to the E. coli GLP-induced RF response, as we have found that macrophages require the expression of MyD88 in order to respond to GLP by nitric oxide production (data not shown). TLR ligation has been shown previously to contribute to the RF response .
Our data are consistent with the findings that RF induction frequently accompanies HCV infections, and indicate that damage to the liver endothelium and induction of hepatocyte apoptosis via Fas may be sufficient to break tolerance and induce RF production, even in the absence of an infectious stimulus. We demonstrate here that sublethal Fas-induced liver injury in vivo caused a breach in the endothelial lining of the liver and led to the loss of self-tolerance.
This work was supported by grants from the Canadian Institutes for Health Research (grant no. 53337 to J. D.).
- 1Liver immunology. Philadelphia: Hanley & Belfus, 2003:15–29, 101., , .
- 4Ceramide, tumor necrosis factor and alcohol-induced liver disease. Alcohol Clin Exp Res 2005; 29 (Suppl 11):S151–7., , , .
- 28The B cell repertoire of patients with rheumatoid arthritis. Frequencies and specificities of peripheral blood B cells reacting with human IgG, human collagens, a mycobacterial heat shock protein and other antigens. Clin Exp Immunol 1993; 92:404–11., , , , .