Parenchymal Expression of CD40 Exacerbates Adenovirus-Induced Hepatitis

Abstract

The healthy adult human liver expresses low levels of MHC II and undetectable levels of immune co-stimulatory molecules. However, high levels of MHC class II, CD40 and B7 family molecules are expressed in the activated Kupffer cells and hepatocytes of patients having viral hepatitis. The precise role of these molecules in viral clearance and immune-mediated liver injury is not well understood. We hypothesize that parenchymal CD40 expression enhances T-cell recruitment and effector functions, which may facilitate viral clearance and alleviate liver injury. To test this hypothesis, we generated novel, liver-specific, conditional CD40 transgenic mice, and challenged them i.v. with recombinant replication-deficient adenovirus carrying Cre recombinase (AdCre). Wild-type mice infected with AdCre developed a relatively mild course of viral hepatitis and recovered spontaneously. CD40 expression in the liver of transgenic animals, however, resulted in CD80 and CD86 expression. Dysregulation of population dynamics and effector functions of intrahepatic lymphocytes results in severe lymphocytic infiltration, apoptosis, necroinflammation, and serum alanine transferase (ALT) elevation in a dose-dependent fashion. To our surprise, an early expansion followed by a contraction of intrahepatic lymphocytes, especially CD8⁺ and NK cells, accompanied by increased granzyme B and IFN-γ production, did not lead to a faster viral clearance in CD40 transgenic mice. Conclusion: Our results demonstrated that hepatic CD40 expression does not accelerate adenoviral clearance, but rather exacerbates liver injury. This study unveils a previously unknown deleterious effect of hepatic CD40 in adenovirus-induced liver inflammation.

Introduction

Adenoviruses are responsible for approximately 5% of all upper respiratory infections and considerable cases of gastroenteritis in the developing world and among immunosuppressive individuals globally. In addition to its role as an important pathogen, recombinant adenovirus, especially serotype 5 (Ad5), is one of the preferred vectors for gene therapy and experimental vaccines for HIV. More than 250 clinical trials of Ad5 were conducted from 1993 to 2007 (www.wiley.co.uk/genmed/clinical). The virus targets the liver, airways and lymphocytes preferentially. However, it can also induce strong T helper (Th), cytotoxic T lymphocyte (CTL) and B cell responses against the viral vector and the transgene in the presence of CD40/CD40 ligand (CD40L) and B7/CD28 co-stimulatory signals (1). Failure to constrain these responses can lead to necroinflammatory hepatitis, treatment failure and even patient death (2). Disruption of the co-stimulatory pathways and immune responses, on the other hand, can enhance adenovirus-mediated gene transfer into the liver (3). The involvement of co-stimulatory pathways in T cell-mediated hepatitis is not peculiar to adenovirus. In hepatitis C virus infection, high levels of MHC class I, II, CD40, and B7 family co-stimulatory molecules were strongly expressed on the activated Kupffer cells and hepatocytes in patient liver, and their levels were closely correlated with those of intrahepatic inflammation, necrosis and elevation of serum alanine aminotransferase (ALT) levels (4–8). Despite these seeming associations, however, the precise role of parenchymal B7 superfamily molecules in viral clearance and liver inflammation is not entirely clear, partly due to severe restrictions in human studies and a general lack of suitable small animal models.

The goal of this study is to examine the role of parenchymal CD40 in the course of adenovirus-induced hepatitis. We previously showed that CD86 expression in the HCV transgenic animals resulted in T cell activation and accumulation in the liver, leading to pronounced hepatic inflammation (9). Based on these observations, we speculate that parenchymal CD40 expression is critical in regulating B7 molecule expression and hepatic inflammation, and also raise the question of whether the host may benefit from hepatic expression of co-stimulatory molecules, for instance, facilitating a faster viral clearance in vivo. To address these possibilities, here we generated novel, liver-specific, conditional CD40 transgenic mice. Upon injection of these animals with adenovirus-carrying Cre DNA recombinase (AdCre), the transgene underwent DNA recombination, resulting in CD40 expression. We report, for the first time, that CD40 expression on hepatocytes mediated a transient surge of intrahepatic lymphocytes (IHLs) accompanied by augmented granzyme B as well as IFN-γ production. The activated intrahepatic T and NK cells did not promote a faster viral clearance, but rather, resulted in more severe liver inflammation.

Materials and Methods

Generation of liver-specific conditional CD40 transgenic mice

A 1.5-kb, loxP-flanked DNA fragment was PCR amplified from a pAlbSVPA-HCV-S-derived construct containing loxP (9). CD40-expressing plasmid (pLIVE-mCD40) was produced by inserting murine CD40 cDNA (a gift from Dr. E. Clark of the University of Washington (10)) into plasmid pLIVE (Mirus Bio LLC, Madison, WI) at the Sac I/Xho I sites. Conditional CD40-expressing plasmid (pLIVE-loxP-mCD40) was constructed by inserting the loxP fragment into pLIVE-mCD40 at the Asc I/Sac I sites. Recombination was induced by infecting adenovirus-encoding Cre recombinase (11) and detected by PCR amplification using the following primer pairs: 5′-ggaaccaatgaaatgcgagg-3′ (forward, P5) and 5′-gcacagccgaggcaaagacacc-3′ (reverse, P6). Transgenic mice were produced by microinjection of a 4.0-kb BglII/SbfI fragment containing the mouse CD40 expression cassette into pronuclei of fertilized eggs of C57BL/6J × C3H mice. Transgene-positive founders were identified by PCR amplification with primers P5 and P6. Cycling conditions were as follows: 94 °C for 45 sec, 58 °C for 60 sec, and 72 °C for 120 sec for 30 cycles. Experiments were performed with 2 lineages of mice having similar levels of CD40 expression. Mice were maintained under specific pathogen-free conditions and housed in a conventional animal facility at The University of Texas Medical Branch (UTMB).

Animal experiments

We used age- and sex-matched CD40-transgenic mice in the C57BL/6J × C3H background and their littermate controls. In most experiments, animals were injected i.v. with 2 × 10⁹ pfu AdCre in 200 μl PBS. Negative control mice were injected with PBS. At day 7 and 14 postinfection, mice were either suffocated by CO₂ without perfusion, or anesthetized by i.p. injection with 50 mg/kg sodium pentobarbital prior to perfusion. For non-perfused mice, blood was withdrawn by heart puncture and serum was obtained. Serum ALT levels were measured in the clinical chemistry laboratory at UTMB. At the same time, mouse liver and spleen tissues were also collected for further analyses. Liver histology, TUNEL assay, immunostaining as well as quantitative PCR assay are described in the Supplemental Materials.

Isolation of hepatocytes and intrahepatic lymphocytes (IHLs)

Hepatocytes from wild-type and transgenic mice were isolated as described by Klaunig et al (12). Briefly, mouse liver was first perfused with HBSS without calcium and magnesium, followed with Hank’s buffer with calcium and magnesium plus collagenase D (Roche Applied Science, Indianapolis, IN). Isolated hepatocytes were suspended in L-15 medium. For IHL isolation, liver tissues were removed and pressed through a 200-gauge stainless steel mesh. The liver cell suspension was collected and suspended in RPMI-1640 medium (HyClone, Logan, UT). Liver mononuclear cells were purified by a density gradient centrifugation in Lympholyte-M (Burlington, NC). Total numbers of IHL per liver were calculated. The relative percentages of CD4⁺, CD8⁺, NK and NKT cells were measured by FACS, and the absolute numbers of these lymphocyte subpopulations per liver were calculated according to their percentages and total IHL numbers in individual livers.

Flow cytometry analysis

The following specific mAbs and their corresponding isotype controls were purchased from BD Pharmingen (San Diego, CA) and eBiosciences (San Diego, CA): PE- conjugated anti-CD40 (3.23) and rat IgG2a; FITC-conjugated anti-IFN-γ (XMG1.2), CD49b (DX5), and rat IgG1 and IgM; PE-conjugated anti-granzyme B (16G6) and rat IgG2b; APC-conjugated anti-CD4 (GK1.5) and rat IgG2b; PE-Cy7 anti-CD8 (53-6.7), and rat IgG2a; and APC-Alexa750-conjugated anti-CD3 (17A2), and rat IgG1. All cell staining procedures were performed on ice. Briefly, cells were blocked with 2% of rat/mouse serum and 1 μg/ml FcRγ blocker (CD16/32), stained for specific surface molecules, fixed/permeabilized with a Cytofix/Cytoperm Kit (BD Biosciences, Franklin Lakes, NJ), and then stained for intracellular molecules. To detect intracellular cytokines, 1 μl/ml of Golgi-plug (BD Biosciences) was added for the last 4 h of cultivation. To detect granzyme B, we performed intracellular staining of freshly isolated IHLs. An Annexin V Apoptosis Detection Kit I (BD Biosciences) was used for T lymphocyte apoptosis analysis. Data were acquired on a FACSCanto (BD Biosciences) and analyzed by FlowJo V8.5 Software (TreeStar, Ashland, OR).

Western blot assay

Proteins were extracted from frozen liver tissues by homogenization with a syringe plunger on ice in a lysis buffer [50 mM Tris, pH 8.0, 150 mM NaCl, 1% NP-40, 1 mM PMSF, and a protease inhibitor cocktail (Sigma)]. After centrifugation at 20,000 × g at 4 °C for 15 min, the supernatant was collected for measuring protein concentration by a Bio-Rad protein assay (Hercules, CA). Equal amounts (70 μg) of proteins were loaded onto 10% SDS- polyacrylamide gels and then transferred to polyvinylidene difluoride membranes (BioRad Laboratories). Membranes were incubated with goat anti-mouse CD40 (clone T-20, Santa Cruz, CA) or anti-β-actin (clone AC-15, Sigma), followed by incubation with HRP-conjugated secondary Abs for 1 h. Blots were visualized by enhanced chemiluminescence (ECL, Amersham Biosciences, Piscataway, NJ).

Statistical analysis

ANOVA analysis was performed. For group-to-group comparison, an unpaired student’s t test was employed. A p-value of less than 0.05 was considered statistically significant.

Results

Generation of liver-specific conditional CD40 transgenic mice

We generated conditional CD40 transgenic mice, which would express CD40 molecules on the surface of hepatocytes only after induction. The CD40 gene was regulated by a chimeric mouse liver promoter, and the two elements were separated by a loxP-flanked DNA spacer, which could be deleted by Cre-mediated recombination (Fig. 1A). Transgenic founders were identified by both PCR and slot blot analyses (Fig. 1B, and data not shown). PCR analysis of the F2 generation from lineage 21 demonstrated a 2.0-kb amplicon, indicative of the unrecombined transgene, while Cre-mediated recombination generated a 0.6-kb DNA fragment (Fig. 1B). Following AdCre transduction, abundant amounts of CD40 mRNA were evident in the liver of transgene-positive, but not negative mice (Fig. 1C). Transgenic mice began to express CD40 in the liver as early as day 3 following AdCre induction, and maintained high levels of transgene expression during the 2 weeks (Fig. 1D and Supp. Fig. 1), which was similar to our previous observations (9). Nearly all hepatocytes in transgenic mice expressed CD40 molecules on their surfaces, as assessed by flow cytometry (Fig. 1E). The transgenic mice were healthy, with normal histology of the liver, spleen, lungs, and kidneys (Supp. Fig. 3C and data not shown), as well as normal liver functions (average ALT 50.4 ± 6.6 U/L).

Figure 1. Transgene construct and characterization of liver-specific conditional CD40 transgenic mice.

(A) A liver-specific conditional CD40 transgene construct and its induction via Cre-mediated recombination. P5 and P6 were the PCR primers, amplifying a 2.0-kb fragment and a 0.6-kb fragment for unrecombined and recombined transgenes, respectively. AdCre, replication-deficient adenovirus carrying Cre recombinase. Note three stop codons in different reading frames in the spacer flanked by loxP. (B) PCR analysis of CD40 transgene recombination in vivo. Transgene-negative and -positive mice from a lineage were i.v. injected with AdCre or PBS. Seven days later, mouse liver genomic DNA was extracted and used for PCR analysis. (C) Quantitative RT-PCR analysis for liver CD40 mRNA. Total liver RNA was extracted from AdCre-infected mice 7 days postinfection. (D) Western blot analysis of CD40 transgene expression. Liver lysates, as described above, were subjected to electrophoresis, and the blot was probed with antibody specific to the C-terminus of murine CD40 (see Materials and Methods). The positive control (lane 1) in this experiment was the liver tissue of a double-transgenic mouse (CD40 × Alb-Cre), which constitutively expressed CD40. Animals were sacrificed at day 7 (lanes 3 and 5) post-AdCre injection. Results were from one representative of two repeated experiments. (E) Flow cytometry analysis for surface CD40 expression on hepatocytes. Mice were i.v. injected with AdCre or PBS, and 5 days later, hepatocytes were obtained from perfused and collagenase-digested livers.

CD40 expression exacerbating liver inflammation in viral hepatitis

To examine the role of CD40 in viral hepatitis, we challenged CD40 transgenic mice i.v. with 2 × 10⁹ pfu AdCre (Tg⁺ AdCre). Two additional groups of wild-type littermates were included as controls, and they were treated similarly with PBS (Tg⁻ PBS) or AdCre (Tg⁻ AdCre). No pathological changes appeared in PBS-treated, wild-type mice, as determined by liver histology and serum ALT levels (Figs. 2 and 3A). AdCre injection in wild-type mice caused liver inflammation characterized by hepatocytes with megaloblastic changes and single-cell necrosis. Sporadic apoptosis and mitosis were observed in the liver at day 7 postinfection (Fig. 2A, arrows). Compared to the control animals, transgenic mice expressed high levels of CD40 in the liver following AdCre injection at 7 and 14 days postinjection and suffered from severe liver injury at day 7 (Supp. Fig. 1 andFig. ). The hepatic inflammation in these Tg⁺ AdCre mice was characterized by prominent portal and lobular lymphocytic infiltration, including CD4⁺, CD8⁺, B220⁺, Mac-1⁺, NK cells and granulocytes (Supp. Fig. 2 and data not shown). Bridging necrosis accompanied by many apoptotic bodies was found in all three adjacent zones at day 7 (Fig. 2A, arrows). At day 14, there were fewer aggregates of lymphocytes in the lobules, and no obvious hepatocyte necrosis or apoptotic bodies were observed (data not shown).

Figure 2. Parenchymal CD40 expression exacerbates liver injury.

Mice were divided into 3 groups (Tg⁻ PBS, tissue from PBS-injected non-transgenic mouse; Tg⁻ AdCre, tissue from AdCre-injected, non-transgenic mouse; and Tg⁺ AdCre, tissue from AdCre-injected, transgenic mouse). The animals were i.v. injected with 2 × 10⁹ pfu AdCre. (A) Liver tissues were obtained from transgenic and non-transgenic mice at day 7 post AdCre injection, and tissue sections were stained with H&E. Arrows denote megaloblastic changes and mitosis in non-transgenic mice, and apoptotic bodies in transgenic animals. TUNEL assay and immunohistochemical analysis were shown in Supp. Fig. 2A and B, respectively. (B) Scores for liver injury. Liver injury was scored with respect to portal and intralobular inflammation, degeneration and necrosis (See Supplemental Materials and Methods). The extent of pathology was scored from 0 (no pathology) to 3 (severe pathology). The liver tissues and serum were obtained at day 7 and day 14 postinfection. The figures summarize the results from two repeated experiments, showing the mean scores of an average of 3 animals in the Tg⁻ PBS group, 3–4 mice in the Tg⁻ AdCre group, and 4–6 mice in the Tg⁺ AdCre group. A significant difference was detected by ANOVA analysis in both day 7 and day 14 groups. The two-tailed t test was used for additional group-to-group comparison. Results were expressed with asterisks (*p < 0.05; **p < 0.01; ***p < 0.001; NS, no significance).

Figure 3. Exacerbated liver injury is not associated with faster viral clearance.

(A) ALT levels of the relative mice in Fig. 2. Results were representative of two experiments. (B) Clearance of intrahepatic AdCre from infected mice. The AdCre genome in the livers of infected non-transgenic and transgenic mice was quantitated by real-time PCR analysis. Each dot represented an individual mouse, and data were pooled from three independent experiments. Results were expressed with asterisks (*p < 0.05; **p < 0.01; ***p < 0.001; NS, no significance).

To quantify the histopathological changes, three individuals scored the results double blindly, and they were then subjected to ANOVA analysis. Our findings demonstrated that viral infection was associated with higher histopathological scores in both transgenic and wild-type animals on day 7 postinfection (p < 0.05, Fig. 2B). Furthermore, CD40 expression resulted in exacerbated liver injuries in transgenic mice, compared with the findings in the infected non-transgenics at day 7 (2.0 ± 0.8 vs. 0.9 ± 0.2, p < 0.05). Despite persistent CD40 expression in the liver (Supp. Fig. 1), the histopathological scores in transgenic mice subsided considerably at day 14. Although the average score in the transgenic group remained slightly higher than that of the infected non-transgenics, the difference between the two groups of animals became statistically insignificant (1.4 vs. 1.0, p > 0.05).

Additional experiments demonstrated that the above-observed pathologic changes were not due to an inherent, unrelated property of the CD40 transgenic mice (Supp. Results). When transgenic mice were i.v. injected with 0, 0.5 ×, 1.5 ×, 2 × and 3 × 10⁹ pfu AdCre, they displayed a dose-dependent CD40-mediated effect on liver inflammation (Supp. Fig. 3). Finally, wild-type adenovirus (wtAd5) and its replication-defective counterpart (AdCre) could also elicit similar viral hepatitis in CD40 transgenic animals (Supp. Fig. 4).

Increased apoptosis and hepatic injury not associated with a faster viral clearance

Apoptosis was long considered to be a natural mechanism of cell removal without pathogenic consequences to the tissue; however, excessive apoptosis can cause tissue injury, and is emerging as an important feature of liver injury (13). By using H&E and TUNEL assays, we found no and low levels of apoptosis in PBS- and AdCre-injected, wild-type mice, respectively, at day 7 postinfection. Many more apoptotic bodies and TUNEL-positive cells were found in the transgenic mice at day 7 (Fig. 2A and Supp. Fig. 2A). The significant morphological difference in initial liver injury between the transgenic and wild-type mice was further confirmed by the measurement of liver injury on day 7, which resulted in an the average serum ALT level of 1,256 U/L for CD40 transgenic mice vs. 263 U/L for wild-type animals (Fig. 3A). By using quantitative PCR analysis, we found no significant difference in viral copy numbers between CD40 transgenic and wild-type groups at day 7 (p > 0.05; Fig. 3B). Although viral copy numbers in both groups decreased steadily from days 7 to 14 (p < 0.01), no statistical difference was found between the two groups at day 14 (p > 0.05). These results demonstrated that increased lymphocyte infiltration and hepatic inflammation were not associated with enhanced viral clearance in the liver.

Effect of parenchymal CD40 expression on lymphocyte recruitment and effector functions

To test how parenchymal CD40 expression exacerbates liver injury in viral hepatitis, we examined population dynamics and effector functions of IHLs in all three groups of mice. As expected, total numbers of IHLs in AdCre-infected mice, irrespective of their transgenic status or point of time, were significantly higher than those in the PBS group (Fig. 4A). The effect of parenchymal CD40 expression on lymphocyte accumulation in the liver was the most evident at day 7, as the average IHLs in transgenic animals rose significantly higher than did those in wild-type animals (29.3 vs. 18.2 × 10⁵, p < 0.01). While the increased IHL numbers were sustained in wild-type mice into the second week (18.5 × 10⁵), the IHL numbers in transgenic animals declined nearly 3-fold to 10.1 × 10⁵, which was significantly lower than those of non-transgenics (p < 0.01).

Figure 4. Apoptosis results in shrinkage of the IHL population in CD40 Tg mice.

Mice were intravenously injected with 2 × 10⁹ pfu AdCre and sacrificed at day 7 and day 14. (A) Intrahepatic lymphocytes at day 7 and day 14 postinfection. (B) Apoptosis in intrahepatic lymphocytes induced by AdCre infection at day 7 postinfection was analyzed with 7-AAD labeling and Annexin V binding on CD8⁺ and CD8⁻ T cells, and shown as mean percentages ± standard error of cells per liver. Data were analyzed for statistical significance by the t test (*p <0.05; **p < 0.01; ***p < 0.001; NS, no significance). Density plots are shown in Supp. Fig. 6.

By using flow cytometry, we found that the adenoviral infection resulted in an increase in the percentages of intrahepatic CD8⁺ cells in both groups of mice at day 7 (57.9% and 62.0%, Table 1), levels higher than that of the PBS group (21.4%; p < 0.001). This CTL expansion was more vigorous in CD40 transgenic mice compared to their wild-type counterparts (18.2 vs. 10.5 × 10⁵), contributing to their more expanded IHL populations (Fig. 4A). Although both AdCre-infected groups maintained high percentages of CD8⁺ T cells in the liver at day 14 (76.3% and 77.5%), transgenic mice had far lower numbers of CD8⁺ cells than wild-type animals due to their greatly diminished IHL pools at day 14 (7.8 vs. 14.1 × 10⁵). Compared to those in wild-type animals, more intrahepatic CD8⁺ cells in CD40 transgenic mice entered the apoptosis process (Annexin V⁺ 7-AAD⁻) as early as day 7 (Supp Fig. 6 and Fig. 4B). This accelerated rate of apoptosis only occurred among CD8⁺ effector cells in transgenic mice, but not in CD8⁻ cells (presumably CD4⁺, B and NK cells). Indeed, the percentages of intrahepatic CD4⁺ T cells in CD40 transgenic mice were relatively comparable to those of wild-type animals, although their numbers were slightly higher and later became lower than those of non-transgenics (Table 1).

TABLE 1.

The percentages and numbers of IHL populations in mice^a

	Tg− PBS	Tg− AdCre	Tg+ AdCre
Day 7
NKT cells	25.4 ± 2.9 (0.9)	7.8 ± 0.9*** (1.4)	7.5 ± 0.8*** (2.2)
NK cells	22.4 ± 2.6 (0.8)	22.4 ± 2.2 (4.1)	21.5± 1.6 (6.3)
CD4+ T cells	51.9 ± 5.5 (1.8)	32.3 ± 3.3* (5.9)	28.6 ± 2.6** (8.4)
CD8+ T cells	21.4 ± 3.2 (0.7)	57.9 ± 4.3*** (10.5)	62.0 ± 2.7*** (18.2)
Day 14
NKT cells	23.9 ± 0.9 (0.8)	3.7± 1.4*** (0.7)	3.5 ± 1.3*** (0.4)
NK cells	19.2 ± 3.4 (0.6)	17.4 ± 2.5 (3.2)	20.0 ± 2.7 (2.0)
CD4+ T cells	48.5 ± 4.5 (1.6)	17.3 ± 1.5*** (3.2)	16.6 ± 1.5*** (1.7)
CD8+ T cells	24.0 ± 5.6 (0.8)	76.3 ± 2.8*** (14.1)	77.5 ± 2.3*** (7.8)

^aThe mean absolute numbers (×10⁵) are shown in parenthesis. Percentages of CD4⁺, CD8⁺ T and NKT cells are gated on CD3⁺ events. Percentages of NK cells are gated on CD3⁻ events.

^*p < 0.05,

^**p < 0.01 and

^***p < 0.001 are compared to Tg⁻ PBS. Data without asterisks indicate p > 0.05.

NK cells are one of the early effector cells in response to adenovirus infection (14, 15). In wild-type mice infected with AdCre, the intrahepatic NK cell population remained relatively stable in both the percentage and absolute numbers between 7 and 14 days postinfection (22.2% to 17.4% and 4.1 to 3.2 × 10⁵, respectively, Table 1). Despite the similarity in their percentages (21.5% and 20.0%), however, the average numbers of intrahepatic NK cells in CD40 transgenic mice were decreased considerably due to the much contracted IHL pools from days 7 to 14 (6.3 to 2.0 × 10⁵, respectively). As other viral infections (16), AdCre infection resulted in a significant decline in NKT percentages compared to those in PBS-injected animals at both days 7 and 14 postinfection (Table 1).

To test whether hepatic CD40 expression modulated phenotypical changes, cytokine production and cytotoxicity in IHLs (14), we measured CD40L levels, intracellular IFN-γ in CD8⁺ and CD4⁺ T cells ex vivo (Supp. Fig. 5 and Fig. 5). The presence of CD40 on hepatocytes did not change the expression of CD40L in CD4⁺ T cells in the liver (Supp. Fig. 5). Following viral infection, higher percentages of CD8⁺ and CD4⁺ T cells in transgenic mice expressed IFN-γ compared to those in non-infected control animals at day 7 (p < 0.05 and p < 0.01, respectively, Supp. Fig. 8A and Fig. 5). Nearly one half of CD4⁺ cells from both transgenic and wild-type mice expressed IFN-γ at day 14 postinfection, and there were, however, no differences between these two groups of cells (Fig. 5). Granzyme B resides in the cytotoxic granules and is a key effector molecule of CD8⁺ CTLs and NK cells (15). To assess the effect of CD40 on granzyme B-mediated target cell destruction, we measured granzyme B-expressing CD8⁺ T cells and NK cells. Similar percentages of CD8⁺ cells in transgenic and control mice expressed granzyme B at day 7 postinfection (Supp. Fig. 8B and Fig. 6). After a phase of increased apoptosis and population contraction (Fig. 4 and Table 1), however, a small percentage CTLs (11.3%) produced greater amounts of this lytic molecule as measured by mean fluorescence intensity (MFI) in transgenic mice at day 14 (p < 0.05; Supp. Fig. 8B and Fig. 6). Following a similar course of population changes (Fig. 4A, and Table 1), higher percentages of intrahepatic NK cells in transgenic mice secreted granzyme B at day 14 (p < 0.05; Supp. Fig. 8B and Fig. 6). Overall, these results suggested that parenchymal CD40 expression can perturb the population dynamics of CTL and NK cells in the liver and alter their effector functions in adenoviral infection.

Figure 5. CD40 expression on hepatocytes increases IFN-γ production by intrahepatic CD4⁺ and CD8⁺ T cells.

Lymphocytes were obtained from perfused livers of non-transgenic and transgenic mice at 7 or 14 days after injection with PBS or AdCre. Following 4 h stimulation with PMA and ionomycin in the presence of Golgi-plug, IFN-γ production was assessed by intracellular staining and FACS analysis. Representative data from a mouse from each group depicting CD8⁺ and CD4⁺ T cells are shown in Supp. Fig. 8A. Shown here is the comparison of the percentages of IFN-γ-producing CD4⁺ and CD8⁺ T cells among three groups. Each group contained 4–6 mice, and the Tg⁻ PBS group was pooled from days 7 and 14 animals. * (p < 0.05) and ** (p < 0.01) indicate statistically significant differences against the Tg⁻ PBS group, and ## (p < 0.01) between the transgenic-positive and – negative groups by the t test. Data without asterisks or # indicate p > 0.05.

Figure 6. CD40 expression on hepatocytes increases granzyme B production by intrahepatic CD8⁺ T and NK cells.

Lymphocytes were obtained from perfused livers of non-transgenic and transgenic mice at 7 or 14 days after injection with PBS or AdCre. Granzyme B production was immediately assessed by intracellular staining and FACS analysis. Representative data from a mouse in each group depicting CD8⁺ T cells and NK cells are shown in Supp. Fig. 8B. Shown here is the comparison of the percentages (upper panels) and MFI (lower panels) of granzyme B expression by CD8⁺ T and NK cells among groups of mice. Each group contained 4–6 mice, and the Tg⁻ PBS group was pooled from days 7 and 14 animals. * (p < 0.05) and ** (p < 0.01) indicate statistically significant differences against the Tg⁻ PBS group, and # (p < 0.05) between the transgenic-positive and -negative groups by the t test. Data without asterisks or # indicate p > 0.05.

Kupffer cells are the resident macrophages in the liver, closely situated adjacent to their neighboring hepatocytes. To test whether CD40 expression on hepatocytes affects Kupffer cells, we measured CD40L levels on Kupffer cells by flow cytometry ex vivo (Supp. Fig. 7). On day 3, Kupffer cells expressed higher levels of surface CD40L than did the later stage (day 12) in both groups. Furthermore, Kupffer cells from CD40⁺ transgenic mice had higher levels of CD40L than those from the control animals on day 3. These data suggest that CD40 expressed on hepatocytes can activate Kupffer cells in the early stage of adenovirus infection. The full implication of this interaction, however, requires further investigation.

B7 family molecules in CD40-transgenic mice

Hepatic CD86 expression is associated with increased T cell activation and retention, contributing to hepatitis in mice (9). In an attempt to test whether parenchymal CD40 expression affects the regulation of B7 family members in the liver, we used qRT-PCR and flow cytometry analyses to examined CD80 and CD86 molecules in transgenic mice 7 days following AdCre injection. CD40 transgenic mice displayed a 1.63-and 1.82-fold increase in CD80 and CD86 mRNA, respectively, over their wild-type littermates (Fig. 7A and B), although the differences were not statistically significant. Furthermore, purified hepatocytes from transgenic mice expressed detectable surface expression of CD80 and CD86 (Fig. 7C to E). The effect of parenchymal CD40 expression was not limited to these two molecules in the B7 superfamily (17); in transgenic mice, the relative copy numbers of PD-L1 (B7-H1) and B7-H4 mRNA were 2.71- (p < 0.01) and 1.84-fold (p > 0.05), respectively, over those in non-transgenics (Supp. Fig. 9). Blocking the PD-1/PD-L1 pathway with anti-PD-L1 antibody further enhanced proliferation, but not IFN-γ expression, of intrahepatic CD8⁺ T cells (Supp. Fig. 10). Consistent with several previous reports (18), mRNA levels of several adhesion molecules, especially E-selectin, also appeared to be upregulated in CD40 transgenic mice (Supp. Fig. 9). These data suggest the possible involvement of B7 family members and adhesion molecules in the pathogenesis of adenovirus-induced hepatitis.

Figure 7. CD40 expression on hepatocytes upregulates downstream co-stimulatory molecules.

(A and B) Total liver RNA was extracted from PBS- and AdCre (2 × 10⁹ pfu)-injected mice at 7 days postinfection. Quantitative RT-PCR assays were carried out for CD80 (A) and CD86 (B). Fold of increase was calculated by normalization to 18s RNA and relative to the CD80 or CD86 quantity in the Tg⁻ PBS group (p = 0.06 and 0.07 by ANOVA, respectively). Each group contained 2–4 animals, and all samples were assayed in triplicate. (C, D and E) Hepatocytes were purified from PBS- and AdCre (2 × 10⁹ pfu)-injected mice at 5 days postinfection. Hepatocytes were stained for surface expression of CD80 and CD86, and data were acquired with flow cytometry. Each group contained 2–4 animals. (C) A representative histogram of CD80 and CD86 was shown in a Tg⁺ AdCre mouse. (D and E) Average MFI for CD80 and CD86 were shown, respectively.

Discussion

CD40 is a member of the TNF-receptor superfamily and is expressed on the surface of professional APCs as well as that of vascular endothelial cells and parenchymal cells during inflammation (4–7). Binding of CD40 by CD40L induces upregulation of MHC and B7 family members on professional APCs, leading to a broad range of immune and inflammatory responses (7, 19, 20). CD40 engagement on vascular endothelial cells induces cell proliferation and expression of adhesion molecules (e.g. E-selectin, VCAM-1 and ICAM-1) (19, 21), resulting in the microvasculature changes in inflammatory bowel disease (22). On primary human hepatocytes, cross-linking CD40 with a mAb leads to a rise in NF-κB and AP-1 activity and apoptosis (19). To date, the role of CD40 in the liver parenchyma in the virus- and immune-mediated hepatitis is not entirely clear, and remains one of the obstacles to gene therapies and orthopedic liver transplantations (2, 23, 24).

The liver is a functionally unique organ, in which hepatic sinusoids allow circulating lymphocytes to make direct contact with underlying hepatocytes through perforated fenestrations of liver sinusoidal endothelial cells (25). Such interactions have been revealed by electron microscopy (26), and ample evidence supports the contention that hepatocytes can act as APCs to direct T cell activation (27–29). We previously reported that hepatic CD86 expression led to hepatitis through T cell activation and accumulation, and speculated that CD40 expression is essential in signaling B7 molecule expression and downstream effects in the liver (9). In this study, we generated transgenic mice that conditionally express CD40 on hepatocytes. Parenchymal CD40 expression upon AdCre infection resulted in an increased expression of CD80 and CD86 molecules, which led to an early expansion followed by a contraction of CD8⁺ T cells in the liver (Table 1). Intrahepatic NK and CD4⁺ cells in CD40 transgenic mice, though to a lesser degree, followed a similar course of population changes, and produced greater amounts of granzyme B and IFN-γ, respectively (Table 1, Figs. 5 and 6). These data unveil that activation of the parenchymal CD40 and B7 signaling pathway disrupts intrahepatic lymphocyte regulation, leading to necroinflammation and severe liver injury. Previously reports have indicated a role of NK cells and CD8⁺ CTLs in different stages of adenovirus infection (14, 15, 30). Dysregulation of intrahepatic lymphocytes can also play a role in other acute and chronic inflammatory liver diseases (4–8, 31).

CD8⁺ CTLs and NK cells are capable of migrating to the liver to produce IFN-γ or degranulating, leading to viral clearance (14, 15, 32). In this study, despite vigorous CD8⁺ T and NK cell responses (Figs. 5 and 6), CD40 transgenic mice did not show an enhanced viral clearance in vivo. In a study designed to dissect the effector functions of virus-specific CTL, the primary CTL clones were reported to produce IFN-γ (cytokine production) or degranulate (cytotoxicity), depending on the antigen concentration (33). Cytotoxicity can be triggered at antigenic peptide concentrations of 10- to 100-fold less than those required for IFN-γ production (33). Indeed, most HBV and HCV were found to be purged from the liver by a cytokine-mediated, non-cytolytic mechanism rather than by direct target destruction (34). Adenovirus-induced hepatotoxicity was linked to granzyme B- and perforin-producing NK cells and CTL (15, 35). More interestingly, a recent study suggested that CD4⁺ T cells can mediate tissue inflammation and liver injury via CD154 in the presence of CD40 (36). Whether this notion is applicable to this and other viral and immune-mediated hepatitis requires further investigation.

Patients with chronic necroinflammatory liver disease had increased percentages of PD-1⁺ intrahepatic lymphocytes, and their hepatocytes expressed its ligands, PD-L1 and B7-DC (8). However, the PD-1/PD-L1 pathway seemed not to affect the acute viral hepatitis in our model (Supp. Fig. 10). In mice, disruption of co-stimulatory molecule PD-L1 resulted in impaired CD8⁺ T cell contraction, leading to accelerated hepatocyte damage and hepatitis (37). In co-stimulatory signaling pathways, CD40 is located upstream of CD80 and CD86; however, whether it interacts with other molecules, including PD-L1, B7-H4 and E-selectin remains unclear (17).

In summary, we generated a novel transgenic mouse model that allows parenchymal CD40 expression after adenovirus infection in the liver. Our results suggested that hepatocyte CD40 expression and activation of its downstream signaling events altered the effector functions of intrahepatic lymphocytes, and exacerbated the liver injury. These data highlight a previously unknown deleterious effect of CD40 engagement and signaling in vivo. These CD40 transgenic mice also provide a valuable model for investigating the relevance of CD40 as the second hit in oxidative stress and altered homeostasis of lymphocytes in alcoholic liver disease and alcoholic steatohepatitis (20, 38).

Abbreviations

ALT

alanine aminotransferase

APC

antigen-presenting cells

CMV

cytomegalovirus

CTL

cytotoxic T lymphocytes

HAI

histology activity index

HCV

hepatitis C virus

IFN

interferon

IHL

intrahepatic lymphocyte

IL

interleukin

mAb

monoclonal antibody

MFI

mean fluorescence intensity

MHC

major histocompatibility complex

TCR

T cell receptor