B-cell depletion with anti-CD20 ameliorates autoimmune cholangitis but exacerbates colitis in transforming growth factor-β receptor II dominant negative mice


  • Potential conflict of interest: Drs. Dunn and Kehry own stock in Biogen Idec.


The treatment of primary biliary cirrhosis (PBC) with conventional immunosuppressive drugs has been relatively disappointing and there have been few efforts in defining a role for the newer biological agents useful in rheumatoid arthritis and other systemic autoimmune diseases. In this study we took advantage of transforming growth factor-β (TGF-β) receptor II dominant negative (dnTGF-βRII) mice, a mouse model of autoimmune cholangitis, to address the therapeutic efficacy of B-cell depletion using anti-CD20. Mice were treated at either 4-6 weeks of age or beginning at 20-22 weeks of age with intraperitoneal injections of anti-CD20 every 2 weeks. We quantitated B-cell levels in all mice as well as antimitochondrial antibodies (AMA), serum and hepatic levels of proinflammatory cytokines, and histopathology of liver and colon. In mice whose treatment was initiated at 4-6 weeks of age, anti-CD20 therapy demonstrated a significantly lower incidence of liver inflammation associated with reduced numbers of activated hepatic CD8+ T cells. However, colon inflammation was exacerbated. In contrast, in mice treated at 20-22 weeks of age, anti-CD20 therapy had relatively little effect on either liver or colon disease. As expected, all treated animals had reduced levels of B cells, absence of AMA, and increased levels in sera of tumor necrosis factor alpha (TNF-α), interleukin 6 (IL-6), and chemokine (C-C motif) ligand (CCL2) (monocyte chemoattractant protein 1 [MCP-1]). Conclusion: These data suggest potential usage of anti-CD20 in early PBC resistant to other modalities, but raise a cautionary note regarding the use of anti-CD20 in inflammatory bowel disease. (HEPATOLOGY 2009.)

The destruction of biliary epithelial cells (BEC) in primary biliary cirrhosis (PBC) is primarily attributed to autoreactive T cells.1–8 In contrast, the contribution of B cells to PBC immunopathology remains unclear,9 despite the nearly universal occurrence in serum of autoantibodies to the pyruvate dehydrogenase E subunit (PDC-E2, antimitochondrial antibodies [AMA]), the major autoantigen in human PBC. Nonetheless, a role for B cells has been suggested. For example, the inflammatory liver infiltrates include foci of B cells.10 Autoantibodies to the E2 subunit of the PDC enzymes inhibit catalytic activity,11, 12 which is reasoned to facilitate immunoglobulin A (IgA)-AMA transcytosis through the BEC in the form of dimeric IgA-AMA complexes leading to the induction of apoptosis of these cells.13 Moreover, autoantibodies to PDC-E2 markedly enhanced the cross-presentation and generation of PDC-E2-specific cytotoxic T-cell responses in the presence of PDC-E2-pulsed antigen-presenting cells.14 However, neither the presence nor levels of AMA following orthotopic liver transplantation for PBC correlate with the recurrence of PBC.15 Thus, notwithstanding the evidence for a profound loss of both B-cell (and T-cell) tolerance to the autoantigenic epitope(s) of PDC-E2, the degree to which B cells or autoantibodies are involved as effector elements in the pathogenesis of BEC damage in PBC remains unresolved.

In one of our mouse models of PBC, the TGF-β receptor II dominant negative (dnTGF-βRII) mice, there is a readily detectable inflammatory lymphocytic infiltrate in the liver16 that closely simulates that seen in chronic nonsuppurative destructive cholangitis of human PBC. In this model, the expression of dnTGF-βRII receptor is restricted to select cell lineages including the CD4+, CD8+, and the CD1d-restricted natural killer T (NKT) cell lineages.16 In efforts to address the role of B cells we treated 4-6-week and 20-22-week-old dnTGF-βRII mice with anti-CD20 monoclonal antibody (mAb). Whereas the PBC-like disease is markedly attenuated in young mice, the colitis was exacerbated. The same therapy of older mice showed no such difference in the severity of either cholangitis and/or colitis. Although anti-CD20 therapy has potential efficacy in PBC, our data argue that effectiveness may be restricted to early-stage patients, and further that there is a potential for exacerbation of any underlying inflammatory bowel disease (IBD).


AMA, antimitochondrial autoantibodies; BEC, biliary epithelial cells; Bregs, B regulatory cells; dnTGF-βRII mice, TGF-β receptor II dominant negative mice; NKT, natural killer T; PBC, primary biliary cirrhosis; TGF-β, transforming growth factor-β.

Materials and Methods

dnTGF-βRII Mice.

dnTGF-βRII mice were bred onto a C57BL/6 (B6) strain background (Jackson Laboratory, Bar Harbor, ME) at the animal facilities of the University of California at Davis.16, 17 Male heterozygous dnTGF-βRII mice were bred with female B6 mice to obtain female heterozygous dnTGF-βRII mice, which were genotyped to confirm the dnTGF-βRII gene in their genomic DNA by the detection of the CD4 promoter at the age of 3-4 weeks.17 All mice were fed a sterile rodent Helicobacter Medicated Dosing System (three-drug combination) diet (Bio-Serv, Frenchtown, NJ) and maintained in individually ventilated cages under specific pathogen-free conditions.


Anti-mouse CD20 IgG2a and an isotype-matched control mAb utilized herein have been described.18 An optimal dose of sterile antimouse CD20 IgG2a or isotype-matched control mAbs (250 μg / 250 μL in phosphate-buffered saline [PBS]) were injected intraperitoneally into dnTGF-βRII mice using a 27G needle every 2 weeks from 4-6 or 20-22 weeks of age. Peripheral blood samples from individual mice were obtained from the retro-orbital vascular plexus before initial treatment and then at 2-week intervals. Sera were stored at −70°C until use. The Animal Care and Use Committee in University of California Davis approved all studies.

Humoral Immunity.

Serum levels of IgG, IgA, and IgM were measured using a mouse Ig isotype quantitative enzyme-linked immunosorbent assay (ELISA) kit (BETHYL, Montgomery, TX). Known standards were used throughout. Serum levels of anti-PDC-E2 were quantified using an ELISA. Briefly, 96-well ELISA plates were coated with purified recombinant (r) PDC-E2 at 10 μg/mL in carbonate buffer (pH 9.6) at 4°C overnight, washed five times with PBS containing 0.05% Tween-20 (Fisher Biotech, Fair Lawn, NJ) (PBS-T), and blocked with 3% skim milk in PBS for 30 minutes. 100 μL of the sera to be tested diluted 1:200 were added to individual wells of this microtiter plate for 1 hour at room temperature (RT) and the plates rewashed. 100 μL of horseradish peroxidase (HRP)-conjugated antimouse immunoglobulin (G+A+M) (H+L) (1:2,000) (Zymed, San Francisco, CA) was added to each well for 1 hour at RT, and the microtiter wells were rewashed. Immunoreactivity was detected by measuring the optical density (OD) at 450 nm after exposure for 15 minutes to 100 μL of TMB peroxidase substrate (KPL, Gaithersburg, MD). Previously calibrated positive and negative standards were included with each assay.

Flow Cytometry.

Peripheral blood mononuclear cells (PBMC) were isolated from heparinized murine blood using Accupaque (Accurate Chemical & Scientific Company, Westbury, CT) to confirm the levels of B cells. Cells were preincubated with antimouse FcR blocking reagent and then incubated at 4°C with a predetermined optimum concentration of PE-Cy5.5-conjugated anti-CD4 (BioLegend), PE-conjugated antimouse IgM (Caltag), and FITC-conjugated anti-CD19 (BioLegend). In addition, mononuclear cells were isolated from liver and spleen suspensions as described.16 An aliquot of these cells was preincubated with antimouse FcR blocking reagent and then incubated at 4°C with a combination of fluorochrome-conjugated Abs, including PE-conjugated anti-CD4 (Pharmingen, San Diego, CA), PE-Cy5 conjugated anti-CD8a (eBioscience), APC-Cy7-conjugated anti-CD3 (BD Biosciences, San Jose, CA), APC-conjugated anti-NK1.1 (BioLegend), FITC-conjugated anti-CD44 (BioLegend), and PE-Cy5-conjugated anti-CD19 (BioLegend). Multicolor flow analyses were performed using a FACScan flow cytometer (BD Immunocytometry Systems, San Jose, CA) upgraded by Cytec Development (Fremont, CA) to allow for five-color analysis. Acquired data were analyzed with CELLQUEST Software (BD Biosciences).

Liver, Spleen, Ileum, and Colon Tissue Preparations.

Portions of the liver, spleen, terminal ileum, and colon were excised and immediately fixed with 10% buffered formalin solution for 2 days at RT. Paraffin-embedded tissue sections were then cut into 4-μm slices for routine hematoxylin (DakoCytomation, Carpinteria, CA) and eosin (American Master Tech Scientific, Lodi, CA) (H&E) staining. Scoring of liver inflammation was performed on coded H&E-stained sections of liver using a set of three indices by a “blinded” pathologist (K.T.); indices included degrees of portal inflammation, parenchymal inflammation, and bile duct damage were scored as 0 = nil, 1 = minimal, 2 = mild, 3 = moderate, and 4 = severe pathology. Scoring of ileum and colon inflammation was based on four indices that included degrees of mucosal erosion, inflammation, epithelial change, and presence of crypt abscesses.

Cytokine Analysis.

The inflammatory cytokines tumor necrosis factor alpha (TNF-α), interleukin 6 (IL-6), IL-10, IL-12p70, and IFN-γ, and the chemokine (C-C motif) ligand (CCL2) (monocyte chemoattractant protein 1 [MCP-1]) were quantitated in sera and in preparations of total liver protein using a mouse cytokine cytometric bead array (BD Biosciences). To extract total liver protein, 100-200 mg of frozen liver tissue were homogenized in TNE buffer (1% Nonidet-P40, 10 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA) containing a cocktail of protease and phosphatase inhibitors (Roche, Indianapolis, IN), as described.19 The suspension was centrifuged at 7,200g for 20 minutes at 4°C and stored at −80°C. The protein concentration was measured using the BCA kit (Pierce Biotechnology, Rockford, IL). Fifty μg of this protein preparation was utilized for the quantitation of cytokines.


Abs against CK22 (monoclonal mouse antihuman cytokeratin (crossreactive with murine cytokeratin; GeneTex, San Antonio, TX), TNF-α (rabbit antimouse TNF-α, 1:20, Monosan, Uden, The Netherlands), IL-6 (IL-6 goat-polyclonal antimouse IL-6, 1:100, R&D Systems, Minneapolis, MN), and MCP-1 (goat-polyclonal anti-MCP-1, 1:100, Santa Cruz Biotechnology, Santa Cruz, CA), were used for immunohistochemical staining of liver and colon sections as previously described.20 Tissue sections were cut at 4μm from tissue blocks and placed on slides. After deparaffinization, sections were soaked in target retrieval buffered saline (TRS, pH 6.1, DakoCytomation) in a nonmetal-containing plastic-made pressure cooker and irradiated in a microwave oven for 10 minutes (maximum 500W). After irradiation, sections were rinsed under running water for 2 minutes, soaked in 3% H2O2 methanol solution for 5 minutes, and then soaked in 5% BSA for 1 minute. Primary antibodies were diluted to a previously determined optimal concentration in PBS containing 5% BSA. The diluted antibodies were applied to the tissue sections in a moist chamber and irradiated intermittently for 10 minutes (250W, 4 seconds-on, 3 seconds-off). After three washes with Tris-buffered saline containing 1% Tween (TBS-T) for 1 minute, peroxidase-conjugated Envision kit for rabbit (Envision-PO, Envision System, DakoCytomation), Histofine-PO for goat and mouse (Nichirei, Tokyo, Japan) were applied on the appropriate specimens in the moist chamber. Irradiation was then performed intermittently for 10 minutes, as described above. After washing 5× with TBS-T, the sections were immersed in DAB solution (Sigma-Aldrich) with H2O2 and counterstained with hematoxylin (DakoCytomation) and mounted under coverslips.

Presentation of Data and Statistical Analysis.

Values are expressed graphically as the mean ± standard error of the mean (SEM). Differences were tested for significance by a two-tailed unpaired Mann-Whitney test. The frequency of portal inflammation and bile duct damage were evaluated using Fisher's exact test. Values having P < 0.05 were considered statistically significant.


Anti-CD20 Treatment Depletes Peripheral B Cells, AMA, and Sera Immunoglobulin in dnTGF-βRII Mice.

We monitored the depletion of peripheral CD19+ B cells following the intraperitoneal administration of anti-CD20 mAb. As noted in Fig. 1A,B, the frequency of CD19+ B cells decreased to nearly undetectable levels at 2 weeks after the initial treatment at 4-6-week and 20-22-week-old dnTGF-βRII mice. Sera from dnTGF-βRII mice contained significant levels of anti-rPDC-E2 prior to treatment with the anti-CD20 antibody. However, these titers decreased markedly following treatment (Fig. 2). This decrease in rPDC-E2 reactivity was accompanied by a significant reduction of serum IgM, IgA, and IgG (Fig. 2).

Figure 1.

Time course of peripheral frequency of CD19+ cells in anti-CD20-treated dnTGF-βRII mice. (A) Representative lymphocyte gated plots are shown for before and 2 weeks after anti-CD20 treatment. (B) Administration of anti-CD20 mAb into peritoneal cavity from 4-6 weeks and 20-22 weeks of age efficiently deprived CD19+ B cells from peripheral blood in dnTGF-βRII mice 2 weeks after initial treatment. Frequency of CD19+ B cells was significantly lower in anti-CD20-treated dnTGF-βRII mice (n = 7 and 7 in younger and older groups, respectively) than controls (n = 9 and 7). (***P < 0.001 in Mann-Whitney test in B.)

Figure 2.

Serum levels of anti-PDC-E2 and immunoglobulin in anti-CD20-treated dnTGF-βRII mice. Serum reactivity to PDC-E2 was significantly lower in anti-CD20-treated dnTGF-βRII mice (n = 6 and 5 in younger and older groups, respectively) than controls (n = 6 and 5) 4 weeks after initial treatment, whereas the reactivity was higher in mice assigned for B-cell depletion in the younger group. Serum levels of IgM, IgA, and IgG were also significantly reduced by anti-CD20 treatment. (*P < 0.05, **P < 0.01 in Mann-Whitney test.)

Anti-CD20 Treatment Ameliorates PBC-like Liver Disease in Young dnTGF-βRII Mice.

Liver sections from anti-CD20-treated dnTGF-βRII mice demonstrated a marked diminution of liver inflammation and bile duct damage as determined by the use of anti-CK22 to highlight hepatocytes and cholangiocytes (Fig. 3A). In addition, hepatic inflammatory cell infiltrates were seldom observed in comparison to dnTGF-βRII mice administered the control mAb. Thus, anti-CD20 treatment ameliorated liver inflammation and bile duct damage in dnTGF-βRII mice (Fig. 3B,C). The degrees of portal tract and hepatic parenchymal inflammation plotted individually are shown in Fig. 3D. Of note, the fortnightly treatment regimen of the anti-CD20 mAb described above did not induce any inflammation in the liver or colon of control B6 littermates (data not shown).

Figure 3.

(A) PBC-like liver pathology was ameliorated in anti-CD20-mAb-treated younger, but not older, dnTGF-βRII mice. Liver sections at 20-22 weeks of age after 16 weeks treatment showing in anti-CD20-mAb-treated dnTGF-βRII mice milder levels of cellular infiltrates around interlobular bile ducts than in control-mAb-treated dnTGF-βRII mice. CK22 demonstrated damaged bile ducts in control dnTGF-βRII mice. (B) Liver inflammation was evaluated in each liver section. Frequency of liver inflammation-positive sections was significantly lower in anti-CD20-Ab (n = 7) -treated dnTGF-βRII mice than control-mAb-treated mice (n = 9). (C) Bile duct damage was also evaluated for each sample. (D) Liver, portal, and parenchymal inflammation score was demonstrated for each sample in the younger group of mice. (E) Some livers from anti-CD20-treated mice at 36-38 weeks of age after 16 weeks treatment demonstrated milder levels of cellular infiltrates than those of controls. CK22 demonstrated damaged bile ducts in anti-CD20-mAb-treated older mice and controls. Bile duct paucity was demonstrated in a portal area of controls. (F) Frequency of liver inflammation-positive sections did not differ significantly between anti-CD20-treated older dnTGF-βRII mice (n = 7) and controls (n = 7). (G) Bile duct damage was not regulated by anti-CD20 treatment. (H) Liver, portal, and parenchymal inflammation score was demonstrated for each sample in the older group mice. (H&E and CK22 staining. Black scale bars = 100 μm in A and E. *P < 0.05 in Fisher's exact test.)

Anti-CD20 Has No Effect on Older dnTGF-βRII Mice.

The data obtained on tissues from the older (20-22 to 36-38-week-old) dnTGF-βRII mice treated with the anti-CD20 mAb were rather different. Therapy induced neither improvement of liver inflammation nor exacerbation of colitis. In some of the anti-CD20-treated mice, inflammation in the liver was milder and in the colon more severe compared with controls (Figs. 3E-H, 5C,D). Anti-CD20 treatment efficiently depleted B cells in peripheral blood, liver, and spleen (Figs. 1B, 4), and reduced serum reactivity to rPDC-E2 (Fig. 2).

Figure 4.

Immunological profiles of liver mononuclear cells in anti-CD20-mAb- and control-mAb-treated dnTGF-βRII mice. Absolute numbers of CD19+ B, CD4+ T, CD8+ T, CD44hiCD4+ T, and CD44hiCD8+ T cells were demonstrated for gram of liver and spleen. Depletion of CD19+ B cells was accomplished in both liver and spleen by anti-CD20 treatment in younger and older groups. Absolute number of liver CD8+ T cells and their activated phenotypes were significantly decreased in livers of anti-CD20-treated younger dnTGF-βRII mice (n = 6) compared to that of control-mAb-treated mice (n = 8), whereas that difference was not obvious in older mice after anti-CD20 treatment. (*P < 0.05, **P < 0.01, ***P < 0.001 in Mann-Whitney test.)

B-Cell Depletion by Anti-CD20 Reduces the Level of CD8+ T-Cell Infiltrates in the Liver of Young dnTGF-βRII Mice.

Flow cytometric analysis demonstrated that anti-CD20 treatment efficiently depleted B cells in the liver and spleen. Of interest was the finding that, whereas anti-CD20 treatment of dnTGF-βRII mice markedly reduced the numbers of CD8+ T cells and their activated phenotypes in the liver, the frequency of CD8+ T cells were only slightly reduced in spleen and similar to mice treated with the control mAb. On the other hand, the number of CD4+ T cells was comparable in liver of the anti-CD20-treated dnTGF-βRII mice and control mAb treatment (Fig. 4). Anti-CD20 treatment efficiently depleted splenic B cells; however, total and activated cell numbers of CD4+ and CD8+ T lymphocytes in spleen did not differ significantly after anti-CD20 treatment.

Colon Inflammation in Young dnTGF-βRII Mice Exacerbated by Anti-CD20 Treatment.

Histologically, the colon of the anti-CD20-treated dnTGF-βRII mice demonstrated severe inflammation compared with normal B-cell-intact mice (Fig. 5A). The degree of colon inflammation differed significantly for the two groups (Fig. 5B).

Figure 5.

Colitis was exacerbated in anti-CD20-treated younger dnTGF-βRII mice. (A) Sections of colon at 20-22 weeks of age after 16-week treatment showing severe inflammation in all layers in anti-CD20-treated dnTGF-βRII mice contrasting with mild inflammation in dnTGF-βRII mice. (B) Degree of colon inflammation was significantly greater in anti-CD20-treated mice (n = 7) than control mice (n = 9). (C) and (D) Some of anti-CD20-treated colon at 36-38 weeks of age after 16-week treatment demonstrated severer inflammation in anti-CD20-mAb-treated dnTGF-βRII mice contrasting with no inflammation in some of control-mAb-treated colon; however, those did not differ significantly. (H&E staining. Scale bars = 100 μm in A and C, **P < 0.01 in Mann-Whitney test.)

Cytokine Profiles in Serum, Liver, and Colon.

The levels of inflammatory cytokines in dnTGF-βRII mice were modulated by anti-CD20 treatment. Serum levels of inflammatory cytokines, IL-6 and TNF-α, and MCP-1 were greater after anti-CD20 treatment compared to those of similarly treated control mAb-treated mice (Fig. 6A -C), whereas levels of IFN-γ, IL-12p70, and IL-10 were below the levels of detection in many of the serum samples (data not shown). Although the levels of IL-6 were the only proinflammatory cytokine that was significantly increased in extracted total hepatic proteins from the anti-CD20-treated mice as compared with mice treated with the control mAb (Fig. 6A), the levels of IL-12p70, IL-10, and MCP-1 were similar in the anti-CD20 as compared with the mice treated with the control mAb (Fig. 6B and data not shown). TNF-α and IFN-γ were below levels of detection in the liver protein extract of both groups of mice (Fig. 6C and data not shown). The cytokine profile in the liver and colon was also investigated immunohistochemically. IL-6 was readily detected on hepatocytes and cholangiocytes in tissue sections from both the anti-CD20-treated mice and controls (Fig. 6D). In addition, IL-6-positive mononuclear cells were present among colonic cell infiltrates in both groups of mice, and the cytoplasm of epithelial cells on the mucosal surface and within crypt lumens were also positive for IL-6 in the anti-CD20-treated mice. MCP-1 was also readily detectable in hepatocytes and cholangiocytes, and strongly so in Kupffer cells and portal infiltrating lymphocytes. The degree of staining for MCP-1 in infiltrating mononuclear cells was increased in colon tissues of anti-CD20-treated mice compared to controls (Fig. 6E). Although TNF-α-positive cells were observed in the colon, but not in liver tissues of anti-CD20-treated mice, TNF-α-positive cells were lacking in both liver and colon tissues from mice treated with the control mAb (Fig. 6F). IFN-γ was undetectable in either liver or colon tissues irrespective of treatment (data not shown).

Figure 6.

Inflammatory cytokines in anti-CD20-treated younger dnTGF-βRII mice. (A-C) Mean serum cytokine levels in mice at 20-22 weeks of age after 16-week treatment, using a cytometric bead array kit. IL-6, MCP-1, and TNF-α were significantly increased in anti-CD20-treated dnTGF-βRII mice (n = 7) compared with control-Ab-treated dnTGF-βRII mice (n = 9). Levels of IFN-γ, IL-12p70, and IL-10 were lower than detectable ranges in the serum samples from either strain (data not shown). Liver cytokines were also examined in anti-CD20-treated and control mice. Liver IL-6 was significantly higher in anti-CD20-treated mice (n = 3) than controls (n = 4), whereas MCP-1 was comparable and TNF-α was lower than detectable range. (D-F) Immunohistochemical detection of IL-6, MCP-1, and TNF-α. IL-6 was detectable in hepatocytes and bile duct epithelial cells, but faint in mononuclear cells of liver. IL-6-positive mononuclear cells were scattered in colon of both groups, whereas cytoplasm of epithelial cells and lumens of cryptae were IL-6-positive in anti-CD20-treated mice. MCP-1 was detectable in Kupffer cells, mononuclear cells, and hepatocytes, while slightly positive in bile duct epithelial cells in liver. MCP-1 was detected in mononuclear cells in colon. Frequency of MCP-1-positive cells was greater in colon of anti-CD20-treated mice. TNF-α staining demonstrated numerous positive mononuclear cells in colon, but not in liver, of anti-CD20-Ab-treated mice, whereas no positive cells were observed in either colon or liver of control mice. (**P < 0.01, ***P < 0.001 in Mann-Whitney test in A-C for sera. Scale bars = 100 μm in D-F.)


We demonstrate herein that therapeutic deletion of B lymphocytes in the dnTGF-βRII mouse model of PBC using an anti-CD20 Ab ameliorated liver inflammation compared to B-cell-competent dnTGF-βRII mice. Further, liver cell infiltrates from B-cell-depleted dnTGF-βRII mice contained reduced populations of CD8+ T cells and their activated phenotypes, and increased serum levels of TNF-α, IL-6, and CCL2 (MCP-1) compared to B-cell-sufficient dnTGF-βRII mice. Thus, not only do B cells promote the development of PBC-like liver disease in younger dnTGF-βRII mice, but also freshly primed B cells possibly including those with specificity for PDC-E2 may enhance progression of inflammatory liver responses.

Data from our recent report suggests the existence of B-cell subsets with a regulatory function, i.e., B regulatory cells (Bregs) in dnTGF-βRII mice,21 as has been reported for several other autoimmune diseases.22, 23 The dnTGF-βRII mice display features of colitis, in addition to cholangitis,17 and because genetic B-cell depletion in dnTGF-βRII mice augments inflammation in the colon as well as the liver, Bregs are thought to be primarily functional from initiation to progression of both colitis and autoimmune cholangitis in this genetically modified strain. However, others recently demonstrated the existence of B-cell subsets with opposing activities, i.e., promoting or suppressing disease progression. In mouse EAE, the contributions of B cells are dependent on time in the course of disease initiation and progression.24

To examine whether the primary function of B cells differs in the course of disease progression in dnTGF-βRII mice, we studied therapeutic B-cell depletion of dnTGF-βRII mice at younger and older ages in comparison to B-cell–competent dnTGF-βRII mice. To deplete B cells sufficiently, we administered anti-CD20 mAb intraperitoneally.25 Peripheral B cells were well depleted in the two groups 2 weeks after initial treatment using anti-CD20 (Fig. 1).

As described above, whereas B-cell depletion from the 4-week-old to 6-week-old young dnTGF-βRII mice demonstrated a marked reduction of liver disease (Fig. 3), such a reduction of inflammation was not noted in liver tissues from the 20-22-week-old aged mice (Fig. 3). In fact, data from the flow cytometric analysis indicate that the absolute numbers of CD8+ T cells and their activated phenotype of CD44-expressing CD8+ T cells were significantly diminished in the livers of B-cell depleted young, but not old, dnTGF-βRII mice (Fig. 4). Given that CD8+ T cells are the primary contributors to autoimmune cholangitis in our adoptive transfer model,21, 26 and the observed reduction in the CD8+ T-cell population in the liver herein, we can suggest that in young mice B cells mediate the hepatic infiltration of CD8+ T cells from the extrahepatic lymphoid organs.

In addition, B-cell depletion in young dnTGF-βRII mice may suppress development of PDC-E2 primed CD8+ T cells due to the depletion of Ig against PDC-E2, i.e., AMA, which was more efficiently depleted in the sera from the young anti-CD20-treated mice (Fig. 2). Antigen-presenting cells may ligate PDC-E2-Ig immune complexes on their Fc receptors and cross-prime CD8+ T cells into activated PDC-E2 reactive phenotypes.14 Insufficient depletion of Ig specific for PDC-E2 may allow the development of autoreactive T cells in older mice. On the other hand, data herein demonstrate no improvement of colonic disease in mice of any age by anti-CD20 treatment (Fig. 5), which is compatible with anti-CD20 treatment in human IBD.26 Thus, these data suggest that the role of B cells in biliary ductular and colonic inflammatory diseases are distinguishable in young dnTGF-βRII mice.

Because we previously found that serum concentrations of TNF-α and IL-6 were significantly greater in Igμ−/− dnTGF-βRII compared to dnTGF-βRII control littermates,21 we studied an extensive profile of inflammatory cytokines in anti-CD20-treated mice. In the current investigation, serum levels of both of these cytokines, as well as the chemokine CCL2 (MCP-1), were increased upon B-cell depletion. In addition, we examined hepatic protein levels of inflammatory mediators. Although there was a marked increase in the level of IL-6 protein in liver of anti-CD20-treated mice, the levels of CCL2 (MCP-1) were comparable. TNF-α was lower than the detectable range in this assay. However, because there was a discrepancy between serum and hepatic levels of inflammatory cytokines, and to explain the ameliorated liver inflammation, we further investigated cytokine profiles in liver and colon immunohistochemically. In contrast to the increased levels of TNF-α in sera, TNF-α was undetectable in the liver. However, some of the mononuclear cells were positive for TNF-α in the colon of anti-CD20-treated mice (Fig. 6F). Thus, B cells are likely to regulate TNF-α-producing cells in colonic inflammation, which are unlikely to mediate hepatic inflammation in this strain. In fact, anti-TNF-α therapy has been documented to dramatically improve the outcome of human IBD.27 On the other hand, IL-6 was observed in hepatocytes and cholangiocytes in liver and colon tissues from the anti-CD20-treated mice, whereas IL-6 was positive in colon tissues of control mice (Fig. 6D). These data suggest that IL-6 is up-regulated by anti-CD20 treatment of mice but that such increased IL-6 levels play opposing roles in liver and colon inflammation, i.e., participation in the amelioration of the PBC-like liver disease but exacerbation of colitis, as reported in ConA-hepatitis and intestinal inflammation by others.28, 29 Improved liver inflammation is likely to reflect a milder infiltration of CD8+ T cells by B-cell depletion, borne out by flow cytometric analysis showing that the number of CD44-expressing activated CD8+ T cells was significantly decreased in the livers of B-cell-depleted dnTGF-βRII mice. Thus, the amelioration of liver disease in B-cell-depleted mice indicates that the presumed proinflammatory function of hepatic B cells is at least partially mediated by the suppression of IL-6 by activated CD8+ T cells.

Similar to our recent report, we failed to detect IL-10 in the serum of anti-CD20-treated dnTGF-βRII mice and controls (data not shown).21 Also, hepatic IL-10 protein levels were comparable for anti-CD20-treated mice and controls (data not shown). Because anti-CD20-treated mice demonstrated ameliorated liver inflammation compared to controls, it is unlikely that hepatic B-cell-derived IL-10 plays an essential immunoregulatory role in autoimmune cholangitis in this genetically modified strain.

PBC patients frequently demonstrate sicca syndromes,30 and PBC sera have been shown to react with apoptotic blebs and bodies of epithelia of human intrahepatic bile duct as well as salivary gland, but not other cells.31 Immune complexes binding to apoptotic bodies may enhance priming of bile duct and salivary gland reactive CD8+ T cells through antigen presentation.14 Indications for anti-CD20 treatment are expanding in human autoimmune diseases. Rituximab (an antihuman CD20 monoclonal antibody) therapy has shown promise for the treatment of Sjogren's syndrome.32 Also, similar to our present data in a murine model of human PBC model, anti-CD20 treatment ameliorated adenitis of submandibular and lacrimal glands in Id3 knockout mice, a model of primary Sjogren's syndrome.33 In one pilot study, in which rituximab was used for the therapy of PBC patients who were refractory to ursodeoxycholic acid (UDCA), there was reduction of serum alkaline phosphatase, and AMA and IgM levels, accompanied by reduced pruritus and fatigue.34

Again, we demonstrated in the present study that anti-CD20 treatment ameliorated PBC-like liver disease, accompanied with a reduction in hepatic CD8+ T cells in young dnTGF-βRII mice. Because PDC-E2-reactive CD8+ T cells have been considered as main contributors to bile duct damage in human PBC,14 therapeutic B-cell depletion may be efficacious in regulating such damage by suppressing expansion of PDC-E2-reactive CD8+ T cells. Clearly, the mechanism of action of anti-CD20 is complex and our data and that of others argues that it extends far beyond simple B-cell depletion; such observations have been made in several models of autoimmunity as well as in clinical studies of treated patients.35–42 Our approach to address the mechanism in more detail will be to continue adoptive transfer studies, similar to our previous published work.21 Our present data provide a rationale for therapeutic B-cell depletion using rituximab in human PBC, although the timing of therapeutic B-cell depletion may be critically dependent on use in the earlier stages of the disease.


We thank Dr. Katsunori Yoshida and Dr. Guo-Xiang Yang for performing ELISA and for technical support in this experiment. We also thank Ms. Nikki Phipps for support in preparing this article.