High susceptibility to liver injury in IL-27 p28 conditional knockout mice involves intrinsic interferon-γ dysregulation of CD4+ T cells

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Interleukin (IL)-27, a newly discovered IL-12 family cytokine, is composed of p28 and EBI3. In this study, CD11c-p28f/f conditional knockout mice were generated to delete p28 specifically in dendritic cells (DCs). We demonstrated that in the absence of DC-derived p28, these mice were highly susceptible to both low and higher concentrations of concanavalin A (ConA) (5 mg/kg or 10 mg/kg), with extremely early and steady high levels of interferon-γ (IFN-γ) in sera. Neutralizing IFN-γ prevented ConA-induced liver damage in these mice, indicating a critical role of IFN-γ in this pathological process. Interestingly, the main source of the increased IFN-γ in CD11c-p28f/f mice was CD4+ T cells, but not natural killer T (NKT) cells. Depletion of CD4+, but not NK1.1+, cells completely abolished liver damage, whereas transferring CD4+ T cells from CD11c-p28f/f mice, but not from wild-type mice or CD11c-p28f/f-IFN−/− double knockout mice to CD4−/− mice, restored the increased liver damage. Further studies defined higher levels of IFN-γ and T-bet messenger RNA in naïve CD4+ T cells from CD11c-p28f/f mice, and these CD4+ T cells were highly responsive to both low and higher concentrations of anti-CD3, indicating a programmed functional alternation of CD4+ T cells. Conclusion: We provide a unique model for studying the pathology of CD4+ T cell–mediated liver injury and reveal a novel function of DC-derived p28 on ConA-induced fulminant hepatitis through regulation of the intrinsic ability for IFN-γ production by CD4+ T cells. (HEPATOLOGY 2013)

Concanavalin A (ConA)-induced fulminant hepatitis is a well-characterized model for studying the pathogenesis of immunorelated pathology and the underlying mechanisms. 1 Lines of evidence suggest that CD4+ T cells and NKT cells are both critical in this animal model. 1-4 Either the CD1d−/− or Vα14−/− mice are protected from the ConA-induced fulminant hepatitis. 5, 6 Whether there is a situation that it is solely dependent on CD4+ T cells without NKT cells remains unknown. Interferon-γ (IFN-γ) has been demonstrated to play an important role in ConA-induced hepatitis. 7 Both IFN−/− mice and STAT1−/− (signal transducer and activator of transcription 1) mice are very resistant to ConA-induced liver damage. 8, 9 However, the precise mechanisms are still not well defined.

Interleukin (IL)-27 is a heterodimeric cytokine, is composed of p28 (also named IL-27 p28, or IL-30) and EBI3, and is mainly secreted by antigen-presenting cells. 10 As a newly identified member of the IL-12 family cytokines, it has attracted increasing attention in recent years due to its widespread involvement in many kinds of autoimmune diseases and inflammation pathology as both a pro- and anti-inflammation factor. 11-14 Recent studies have revealed that the IL-12 family cytokines (IL-12, IL-23, and IL-27) are involved in ConA-induced hepatitis. 15-17 It has been reported that WSX-1−/− mice (WSX-1, putative IL-27 receptor, also named TCCR) are more susceptible to ConA-induced liver injury than wild-type (WT) mice. 18 In contrast, EBI3−/− mice are protected from severe liver injury upon challenge with ConA. 17 Blockade of IL-27 using soluble WSX-1 ameliorates ConA-induced liver damage, suggesting a key pathogenic role of IL-27 signaling in ConA-induced hepatitis. 17 The apparent contradictions can also be explained by the fact that the p28 subunit may be secreted and biologically active independent of EBI3. 19, 20 A recent study has revealed a protective role of exogenous p28 by using a gene therapy approach. 21 However, the primary source, the pathological role, and the specific mechanism of endogenous p28 in ConA-induced hepatitis are still unknown.

In this report, we investigated the role of dendritic cell (DC)-derived p28 in ConA-induced fulminant hepatitis by generating p28 conditional knockout mice. We demonstrated that in the absence of DC-derived p28, mice were highly susceptible to both low doses and higher doses of ConA, with highly elevated levels of IFN-γ in sera, which appeared earlier and lasted longer than those in WT mice. We then defined that CD4+ T cells, but not NK1.1+ cells, were the primary source of hyper–IFN-γ due to programmed functional alternation of CD4+ T cells. We have thus determined a novel function of p28 in regulating the intrinsic ability of CD4+ T cells to produce IFN-γ, which is critical for the immunopathology in ConA-induced acute hepatitis.

Abbreviations

ALT, alanine aminotransferase; ConA, concanavalin A; DC, dendritic cell; ELISA, enzyme-linked immunosorbent assay; IFN-γ, interferon-γ; IL, interleukin; mRNA, messenger RNA; NK, natural killer; NKT, natural killer T; PCR, polymerase chain reaction; STAT, signal transducer and activator of transcription; WT, wild-type.

Materials and Methods

Mice.

All mice were kept in specific pathogen-free conditions. All animal procedures were approved by the Nankai Laboratory Animal Center. All experiments were performed with age-matched (6-10 weeks) and sex-matched mice (male or female mice were used). Littermate mice with flox/flox, flox/wt, or wt genotype were used as a WT control. C57BL/6J (B6) mice, EIIa-cre mice [B6.FVB-Tg(EIIa-cre)C5379Lmgd/J, stock number 003724], and CD11c-cre mice [C57BL/6J-Tg(Itgax-cre,-EGFP)4097Ach/J, stock number 007567] were purchased from The Jackson Laboratory (Bar Harbor, ME). CD4−/− mice were kindly provided by Dr. Jie Du from An Zhen Hospital (Beijing, China). Rag2-EGFP mice and CD45.1 WT mice were kindly provided by Dr. Yu Zhang from Peking University Health Science Center (Beijing, China). CD11c-p28f/f-IFN−/− double-knockout mice were bred with CD11c-p28f/f and IFN−/− mice. IFN−/− mice were kindly provided by Dr. Richard Flavell from Yale University (New Haven, CT).

Statistics.

Data are presented as the mean ± SEM. Statistical significance between two groups was evaluated using a two-tailed unpaired Student t test. P values are given in the figure legends.

Detailed Materials and Methods can be found in the Supporting Information.

Results

Generation and Characterization of p28 Conditional Knockout Mice.

To facilitate the study on the source and the role of endogenous p28 in immune responses, p28 floxed mice were generated (Fig. 1A). Two LoxP sites were inserted flanking p28 exons 2, 3, and 4 by using the bacteria recombination method (detailed in the Supporting Materials and Methods). The chimeras were bred with C57BL/6J (B6) to obtain B6 p28f/w mice (N11), which were then crossed with B6 CD11c-cre or B6 EIIa-cre mice to obtain CD11c-p28f/f mice as DC-specific deletion mice and EIIa-p28f/f mice as p28-null mice, respectively. Mouse genotyping was performed via polymerase chain reaction (PCR) (Fig. 1B). To confirm the p28 knockout efficiency, p28 expression was analyzed via quantitative real-time PCR (Fig. 1C) and enzyme-linked immunosorbent assay (ELISA) (Fig. 1D). To characterize these mice, the numbers of major immune organ lymphocytes were counted (Supporting Fig. 1A). The composition of immune cells, CD4+ and CD8+ cells (Supporting Fig. 1B), NK and NKT cells (Supporting Fig. 1C), CD4+Foxp3+ Treg cells (Supporting Fig. 1D), and the activation markers CD44 and CD62L (Supporting Fig. 1E) were analyzed. No significant differences were observed between WT and CD11c-p28f/f mice. These mice were healthy and fertile, phenotypically grossly normal, and were born at the expected Mendelian ratio. Therefore, CD11c-p28f/f mice as well as EIIa-p28f/f mice appeared normal in a naïve state.

Figure 1.

Generation and expression analysis of p28 conditional knockout mice. (A) Diagram of p28 conditional knockout mice. Numbers indicate the exons of p28. The p28 locus (top) was targeted by the targeting vector (middle), which contains the homologous sequence of p28, including two LoxP sites flanking exons 2, 3, and 4. Homologous recombination resulted in the floxed allele (bottom). N, NheI; neo, neomycin resistance gene flanked by two FRT sites; S, SalI; tk, thymidine kinase. (B) Genotyping of p28 conditional knockout mice. (C, D) Bone marrow–derived dendritic cells from WT, EIIa-p28f/f, and CD11c-p28f/f mice were (C) cultured and stimulated with 1 μg/mL lipopolysaccharide plus 10 ng/mL IFN-γ for 4 hours and the cells were used for quantitative real-time PCR or (D) stimulated for 24 hours and the supernatant was harvested for ELISA to detect p28 expression (n = 3). **P < 0.01; ***P < 0.001.

EIIa-p28f/f and CD11c-p28f/f Mice Are Highly Susceptible to ConA-Induced Fulminant Hepatitis.

To investigate the role of p28 in ConA-induced fulminant hepatitis, WT, EIIa-p28f/f, and CD11c-p28f/f mice were challenged with different doses of ConA. Surprisingly, even with a very low dose of ConA (5 mg/kg body weight), these EIIa-p28f/f and CD11c-p28f/f mice showed much higher serum alanine aminotransferase (ALT) levels (Fig. 2A) and more severe liver damage (Fig. 2B) than WT mice. About 10%-50% of the knockout mice died due to the fulminant hepatitis. In contrast, this dose brought little liver damage, with very low levels of ALT and no visible histological changes in WT mice. Moreover, CD11c-p28f/f mouse serum ALT levels were strongly induced upon a higher dose (10 mg/kg body weight) of ConA administration (Fig. 2C), and few mice survived. Therefore, we used a low dose (5 mg/kg) of ConA for experiments on CD11c-p28f/f mice in this study. The levels of serum p28 in mice with ConA-induced hepatitis were also analyzed (Supporting Fig. 2). Consistent with the overlap of phenotypes in EIIa-p28f/f and CD11c-p28f/f mice, CD11c-p28f/f mice showed barely detectable serum p28 levels. This finding indicated that DC-derived p28 was playing an important role in ConA-induced fulminant hepatitis as the primary source of circulating p28. We therefore decided to use CD11c-p28f/f mice in subsequent studies.

Figure 2.

EIIa-p28f/f and CD11c-p28f/f mice were highly susceptible to ConA-induced fulminant hepatitis. (A) A total of 5 mg/kg ConA was injected into WT, EIIa-p28f/f, and CD11c-p28f/f mice. Serum ALT levels were measured at 0, 6, 12, 18, 24, 36, 48, and 72 hours after ConA injection (n = 10). (B) Liver tissues were fixed for hematoxylin and eosin staining, and one representative tissue staining is shown. Percentages of necrosis were calculated. N, necrosis area. Scale bars, 200 μm. (C) A total of 10 mg/kg ConA was injected into WT and CD11c-p28f/f mice. Serum ALT levels were measured at 0, 6, 12, 18, 24, 36, 48, and 72 hours after ConA injection (n = 10). Data represent at least three independent experiments with similar results. *P < 0.05; **P < 0.01; ***P < 0.001.

Liver Injury in CD11c-p28f/f Mice Is Dependent on IFN-γ–Mediated Inflammation.

To define the mechanisms of increased susceptibility in CD11c-p28f/f mice, the inflammation cytokine levels were analyzed in sera samples collected at different time points after ConA challenge. A remarkable cytokine storm was observed after a low dose (Fig. 3A) or higher dose (Supporting Fig. 3A) of ConA treatment. Notably, compared with WT mice, an extremely high level of IFN-γ reached its peak at 2 hours, and steadily persisted for 12 hours in CD11c-p28f/f mice. To determine the pathologic role of IFN-γ in this model, CD11c-p28f/f mice were administered with either neutralizing anti–IFN-γ monoclonal antibody or isotype control antibody, and followed by ConA challenge. Interestingly, neutralizing IFN-γ abolished ConA-induced liver damage in CD11c-p28f/f mice, accompanied with much lower levels of ALT and reduced liver necrosis (Fig. 3B,C). To further confirm these results, CD11c-p28f/f mice were crossed with IFN−/− mice to obtain CD11c-p28f/f-IFN−/− double-knockout mice. Indeed, these double-knockout mice were highly resistant to ConA-induced liver injury (Fig. 3D). These results collectively define a critical pathogenic role of IFN-γ in mediating ConA-induced liver injury in CD11c-p28f/f mice.

Figure 3.

Essential role of IFN-γ in mediating liver inflammation in CD11c-p28f/f mice. (A) A total of 5 mg/kg ConA was injected into WT and CD11c-p28f/f mice. Blood sera were collected at 0, 2, 4, 6, 8, 12, and 24 hours after ConA injection. Cytokine levels were measured via ELISA (WT, n = 9; CD11c-p28f/f, n = 14). (B) For neutralizing IFN-γ in vivo, anti–IFN-γ or control antibody were injected intravenously into CD11c-p28f/f mice 2 hours prior to ConA injection, and blood sera were collected for measuring the levels of ALT 18 hours after 5 mg/kg ConA injection (n = 5). (C) Liver tissues were fixed for hematoxylin and eosin staining, and one representative tissue staining is shown. Percentages of necrosis were calculated. N, necrosis area. Scale bars, 200 μm. (D) A total of 5 mg/kg ConA was injected into CD11c-p28f/f mice and CD11c-p28f/f-IFN−/− double-knockout mice. At 18 hours after ConA injection, blood sera were collected and ALT levels were measured (n = 7). (E) A total of 1 mg/kg anti-Fas (clone Jo2) was injected intraperitoneally into WT and CD11c-p28f/f mice. At 12 hours after injection, serum ALT levels were measured (n = 6). The data are representative of three independent experiments with similar results. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.

To exclude the possibility that the increased liver damage in CD11c-p28f/f mice was due to an intrinsic susceptibility to apoptosis independent of inflammation, the response of hepatocytes to activating anti-Fas antibody was evaluated. Injection of mice with this antibody induced hepatocyte apoptosis due to Fas-FasL–mediated killing. WT and CD11c-p28f/f mice showed similar ALT levels (Fig. 3E) as well as liver damage (Supporting Fig. 3B) following this antibody injection. Thus, hepatocytes of these CD11c-p28f/f mice had similar responses to Fas-mediated apoptosis.

CD4+ But Not NK1.1+ Cells Are the Main Source of Increased IFN-γ in CD11c-p28f/f Mice After ConA Treatment.

To elucidate the cellular source for the high level of circulating IFN-γ, WT and CD11c-p28f/f mice were treated with anti-NK1.1, anti-CD4, and control antibodies prior to ConA injection. The cell populations targeted by these antibodies were depleted effectively (Supporting Fig. 4A,B). Depletion of either NK1.1+ cells or CD4+ cells diminished ConA-induced liver damage in WT mice as described (Fig. 4A). 1, 22 Surprisingly, depletion of CD4+ but not NK1.1+ cells dramatically reduced ALT levels and liver damage in CD11c-p28f/f mice (Fig. 4A,C). Moreover, depletion of CD4+ cells but not NK1.1+ cells significantly reduced serum IFN-γ levels (Fig. 4B). These results suggest that CD4+ T cells, but not NK1.1+ cells, were the main source of increased IFN-γ upon ConA administration in CD11c-p28f/f mice.

Figure 4.

CD4+ but not NK1.1+ cells were the main source of increased IFN-γ. (A) For depletion of NK1.1+ cells or CD4+ cells, anti-NK1.1 or anti-CD4 were injected into WT and CD11c-p28f/f mice 24 hours prior to ConA treatment, followed by ConA injection (WT mice, 20 mg/kg; CD11c-p28f/f mice, 5 mg/kg). ALT values of WT and CD11c-p28f/f mice and (B) serum IFN-γ levels of CD11c-p28f/f mice were measured at 12 hours after ConA injection. (n = 5). ND, not detected. (C) Liver tissues from CD11c-p28f/f mice were fixed for hematoxylin and eosin staining, and one representative tissue staining is shown. Percentages of necrosis were calculated. N, necrosis area. Scale bars, 200 μm. (D) A total of 5 mg/kg ConA was injected into WT mice and CD11c-p28f/f mice. Mice were sacrificed, and spleen and liver lymphocytes were isolated at 2 hours after ConA injection. Cells were cultured for 6 hours without any stimulation but in the presence of GolgiStop. IFN-γ expressions in total lymphocytes or (E) in CD4+ T cells (NK1.1CD4+) were analyzed via flow cytometry (n = 8). (F) A total of 5 mg/kg ConA was injected into WT and CD11c-p28f/f mice. At 2 hours after ConA injection, mice were sacrificed, and spleen NK1.1CD4+ cells were purified by magnetic beads (purification efficiency >97%). Cell lysates were analyzed via western blot analysis (eight mice were pooled together for each group). Data represent three independent experiments with similar results. **P < 0.01; ***P < 0.001; ns, not significant.

To verify this conclusion, at 2 hours post ConA injection, spleen and liver lymphocytes from WT and CD11c-p28f/f mice were isolated, cell numbers were counted, CD4+, natural killer (NK) and NKT cell percentages were analyzed, and no significant differences were observed (data not shown). Cells were cultured for 6 hours without any stimulation in the presence of GolgiStop, and intracellular IFN-γ staining was performed. Compared with WT mice, a significantly higher percentage of IFN-γ–producing cells and higher IFN-γ mean fluorescence intensity was observed in the NK1.1 but not NK1.1+ cell population (Fig. 4D). The CD4+ T cells were responsible for the increased cellular IFN-γ (Fig. 4E), supporting our conclusions from the antibody depletion experiments as described. Activation of STATs may mutually regulate each other and tightly control the development and progression of liver diseases. 23 Consistently, these CD4+ T cells from CD11c-p28f/f mice had higher phosphorylation levels of STAT1 and STAT4 (Fig. 4F), indicating the increased IFN-γ signaling was associated with STAT1 and STAT4 in these CD11c-p28f/f mice upon ConA treatment. In summary, our results collectively establish that CD4+ T cells were the main source of increased IFN-γ mediating the pathogenesis of ConA-induced fulminant hepatitis in CD11c-p28f/f mice.

Pathological Role of CD4+ T Cells from CD11c-p28f/f Mice Is IFN-γ–Dependent,

The fact that a large amount of p28 was induced in WT mice (Supporting Fig. 2) after 2 hours when the IFN-γ level had already reached its peak (Fig. 3A) might indicate that p28 was induced by IFN-γ to negatively regulate liver injury, since p28 expression can be regulated by IFN-γ. 24 This is also in agreement with a recent study in which exogenous p28 was shown to be protective in ConA-induced fulminant hepatitis. 21 This suggests a hypothesis whereby the differences between WT and CD11c-p28f/f mice might also be only caused by the deficiency of protective p28 after ConA treatment. To test this hypothesis, we constructed a recombinant mouse p28 plasmid and used the hydrodynamics injection method to express exogenous p28 protein effectively as described (Supporting Fig. 5C), 21, 25 and similar results were obtained. WT mice were protected from ConA-induced fulminant hepatitis (Supporting Fig. 5A), with decreased serum IFN-γ levels (Supporting Fig. 5B), which is consistent with the hypothesis that p28 induced by IFN-γ appears to negatively regulate the liver injury by inhibiting IFN-γ production. Surprisingly, no sign of ameliorated liver injury or serum IFN-γ levels in CD11c-p28f/f mice was observed. These results suggest another hypothesis that CD4+ T cells in CD11c-p28f/f mice were intrinsically changed before administration of ConA.

To test this hypothesis, and to further define whether the pathological role of CD4+ T cells from CD11c-p28f/f mice was IFN-γ–dependent, CD4+ T cells isolated from WT, CD11c-p28f/f, or CD11c-p28f/f-IFN−/− mice were adoptively transferred into CD4−/− mice (Supporting Fig. 6A,B), followed by treatment with ConA (Fig. 5A). Interestingly, only reconstitution with CD4+ T cells from CD11c-p28f/f mice, but not WT mice or CD11c-p28f/f-IFN−/− mice, showed significant pathology as evidenced by ALT levels and histology (Fig. 5B,C). These results suggest that CD4+ T cells from CD11c-p28f/f mice played a pathological role in mediating ConA-induced liver damage through IFN-γ production. Furthermore, the noticeable differences between WT and CD11c-p28f/f mice suggest that CD4+ T cells from CD11c-p28f/f mice are intrinsically different from those of WT mice.

Figure 5.

IFN-γ derived from CD4+ T cells was responsible for the increased susceptibility of CD11c-p28f/f mice. (A) Experimental procedure. A total of 8 million CD4+ T cells from three groups of mice were sorted and transferred into CD4−/− mice. (B) At 18 hours after ConA injection, blood sera were collected and ALT levels were measured (n = 5). (C) Liver tissues were fixed for hematoxylin and eosin staining, and one representative tissue staining is shown. Percentages of necrosis were calculated. N, necrosis area. Scale bars, 200 μm. Data are representative of two independent experiments with similar results. *P < 0.05; **P < 0.01.

CD4+ T Cells in CD11c-p28f/f Mice Have an Intrinsically Elevated Ability to Produce IFN-γ.

To define the functional property of CD4+ T cells from CD11c-p28f/f mice, total spleen lymphocytes from WT and CD11c-p28f/f mice were stimulated in vitro with anti-CD3/anti-CD28 or phorbol 12-myristate 13-acetate/ionomycin (PMA) in the presence of GolgiStop. CD4+ T cells from CD11c-p28f/f mice showed an increased level of IFN-γ production upon both stimulations (Fig. 6A). To further study whether the alternation of the ability for IFN-γ production by CD4+ T cells from the CD11c-p28f/f mice was intrinsic, naïve and memory CD4+ T cells were sorted from both WT and CD11c-p28f/f mice, and directly used for IFN-γ messenger RNA (mRNA) analysis. Interestingly, both the naïve and memory CD4+ T cells from CD11c-p28f/f mice expressed higher levels of IFN-γ mRNA (Fig. 6B,D). Moreover, upon stimulation with a very low level of anti-CD3 (1 μg/mL), naïve CD4+ T cells from CD11c-p28f/f mice were induced to express IFN-γ mRNA (Fig. 6B) and secreted IFN-γ (Fig. 6C). In contrast, naïve CD4+ T cells from WT mice only expressed a low level of IFN-γ with a higher concentration of anti-CD3 (10 μg/mL). These results collectively reveal an intrinsic functional alternation of CD4+ T cells in CD11c-p28f/f mice for their ability of hyper–IFN-γ production.

Figure 6.

CD4+ T cells in CD11c-p28f/f mice had an intrinsically elevated ability of IFN-γ production. (A) Splenocytes from WT and CD11c-p28f/f mice were cultured in medium with GolgiStop in the presence or absence of plate-bound anti-CD3 plus anti-CD28 or PMA plus ionomycin for 4 hours. IFN-γ intracellular staining was performed, and cells were analyzed by flow cytometry. (B) Naïve CD4+ T cells were sorted as CD4+CD44low. A total of 500,000 cells in 200 μL medium were cultured and stimulated with plate-bound anti-CD3 (1 μg/mL or 10 μg/mL, anti-CD3–coated, 96-well plates) and anti-CD28 (1 μg/mL). Cells and supernatant were collected at 6, 12, and 24 hours after stimulation. IFN-γ mRNA levels were determined via quantitative real-time PCR (n = 3). (C) IFN-γ levels in cell supernatant were measured by ELISA (n = 3). (D) Memory CD4+ T cells were sorted as CD4+CD44high. IFN-γ mRNA levels were determined by quantitative real-time PCR (n = 3). Four mice were pooled together for each group, and triplicate samples were used for statistical analysis. Data are representative of at least three independent experiments with similar results. ***P < 0.001.

CD4+ T Cells in CD11c-p28f/f Mice Have Elevated Levels of T-bet.

IFN-γ production has a tight association with T-bet. 26 To further define the underlying mechanisms of the intrinsic ability for IFN-γ production, the expression level of the key transcription factor T-bet was determined. Consistently, naïve CD4+ T cells from CD11c-p28f/f mice expressed a higher level of T-bet mRNA, and the expression level was enhanced upon stimulation with low or higher concentrations of anti-CD3 (Fig. 7A). Higher T-bet levels were also observed in CD4+ T memory cells (Fig. 7B). These findings support the conclusion that CD4+ T cells undergo an intrinsic functional change in CD11c-p28f/f mice.

Figure 7.

CD4+ T cells in CD11c-p28f/f mice had elevated levels of T-bet. (A) Naïve CD4+ T cells were sorted as CD4+CD44low. A total of 500,000 cells in 200 μL medium were cultured and stimulated with plate-bound anti-CD3 (1 μg/mL or 10 μg/mL, anti-CD3–coated, 96-well plates) and anti-CD28 (1 μg/mL). Cells and supernatant were collected at 6, 12, and 24 hours after stimulation. T-bet mRNA levels were determined via quantitative real-time PCR (n = 3). (B) Memory CD4+ T cells were sorted as CD4+CD44high. T-bet mRNA levels were determined via quantitative real-time PCR (n = 3). Four mice were pooled together for each group, and triplicate samples were used for statistical analysis. Data are representative of at least three independent experiments with similar results. ***P < 0.001.

DC-Derived p28 Is Required for Education of CD4+ T Cells in the Thymus.

To determine whether CD4+ T cells were intrinsically changed in the periphery or during CD4+ T cell development in the thymus, we used ectopic transplantation of the thymus under the kidney capsule model to transfer the whole thymus from WT and CD11c-p28f/f mice to congenic host mice (Fig. 8A). 27 CD45.1 and CD45.2 cell surface markers were used to identify the released CD4+ T cells (Fig. 8B). Even with only the thymus transplanted, the CD4+ T cells derived from CD11c-p28f/f mice had an elevated IFN-γ expression isolated from the host spleen (Fig. 8C). These findings indicate the importance of the thymus for the ability of IFN-γ production by CD4+ T cells.

Figure 8.

Thymocyte transplantation. (A) Experimental procedure for ectopic transplantation of the thymus under the kidney capsule. Thymus lobes from CD45.2 WT and CD45.2 CD11c-p28f/f mice were transplanted into CD45.1 WT mice. At 7 days after transplantation, (B) CD45.2+CD4+ T cells were sorted from the host mice spleen. (C) A total of 250,000 cells in 200 μL medium were cultured and stimulated with plate-bound anti-CD3 (1 μg/mL or 10 μg/mL, anti-CD3–coated, 96-well plates) and anti-CD28 (1 μg/mL). IFN-γ mRNA levels were determined via quantitative real-time PCR (n = 3). Nine mice were pooled together for each group, and triplicate samples were used for statistical analysis. (D) Experimental procedure for intrathymic injection. Thymic CD4+CD8+ double-positive cells from Rag2-EGFP mice were sorted and injected into WT and CD11c-p28f/f mouse thymus. At 12 days after injection, (E) GFP+CD4+ T cells were sorted from host mouse spleen and thymus. A total of 150,000 cells in 200 μL medium were cultured and stimulated with 10 μg/mL plate-bound anti-CD3 in 96-well plates. IFN-γ mRNA levels were determined via quantitative real-time PCR (n = 3). Seventeen mice were pooled together for each group, and triplicate samples were used for statistical analysis. ***P < 0.001.

DCs in the thymus are suggested to be critically important for T cell development. 28, 29 To further define whether an intrathymic environment with DCs deficient in p28 can educate CD4+ T cells to produce more IFN-γ, and to define at which stage p28 participates this education, an intrathymic injection approach was performed (Fig. 8D). To identify the newly matured CD4+ T cells, and also to exclude any possible contamination by circulating cells, Rag2-EGFP mice were used as donors to isolate thymic CD4+CD8+ double-positive cells. 30 EGFP is driven by the Rag2 promoter, and lingering EGFP will be detected after T cell maturation (Fig. 8E). After the thymic selection of CD4+ T cells, the injected cells became matured CD4+ single-positive cells and were released to the periphery. The elevated ability for IFN-γ production was also observed in GFP+CD4+ T cells (Fig. 8E). Thus, for the first time, we demonstrate the possibility that DC-derived p28 is important for thymic selection process and intrinsic ability of IFN-γ production by CD4+ T cells.

Discussion

Facilitated by CD11c-p28f/f and EIIa-p28f/f mice we generated, we demonstrated that DC-derived p28 played a critical protective role in vivo in ConA-induced fulminant hepatitis through regulating the intrinsic ability of CD4+ T cells to produce IFN-γ. A ConA-induced fulminant hepatitis model has been used to study the pathogenesis of inflammatory liver damage. 1 IFN-γ and its signaling pathway have been indicated as the critical factor. 7, 8 One of the striking findings from our current study was that in the absence of DC-derived p28, CD4+ T cells were the primary source of pathogenic IFN-γ (Fig. 4). Moreover, IFN-γ derived from CD4+ T cells was the key pathogenic mediator, since CD4+ T cells from CD11c-p28f/f-IFN−/− failed to restore the liver damage of CD4−/− mice (Fig. 5). The more intriguing discovery in these CD11c-p28f/f mice was that the hepatitis was solely mediated by CD4+ T cells even in the absence of NK1.1+ cells (Fig. 4). To the best of our knowledge, this is the first description of a ConA-induced hepatitis model in a CD4+ T cell–mediated but not NK cell– or NKT cell–mediated fashion. Our study thus provides a unique model for studying the molecular mechanisms of CD4+ T cell–mediated liver damage.

Previous studies have demonstrated that NK1.1+ cells are important and responsible for ConA-induced pathology in WT mice. 3-6 Depletion of NK1.1+ cells under different doses of ConA (10 mg/kg, 15 mg/kg, 20 mg/kg) ameliorated hepatitis in WT mice (data not shown). However, in CD11c-p28f/f mice, CD4+ T cells were responsible for severe liver injury by producing large amounts of IFN-γ. Depletion of NK1.1+ cells did not ameliorate hepatitis under low (5 mg/kg; Fig. 4) or high (10 mg/kg, 15 mg/kg; data not shown) doses of ConA. These results indicate that at least in CD11c-p28f/f mice, CD4+ T cells responsible for liver injury were playing the same role under different doses of ConA treatment.

Previous studies have confirmed that IL-27 can promote IFN-γ production of naïve CD4 cells. 31, 32 This is consistent with the result that blockade of IL-27 ameliorates ConA-induced liver damage. 17 Considering that p28 itself can protect the liver from injury, 21 the other subunit of IL-27, EBI3, may be effective against p28 or as a component of IL27 to aggravate liver injury, which is consistent with the observation that EBI3−/− mice were protected from ConA-induced hepatitis. 17 However, different from these short-term functions in the ConA-induced fulminant hepatitis pathological process, CD4+ T cells played a key pathogenic role through IFN-γ production in CD11c-p28f/f mice. We demonstrated that DC-derived p28 affected the ability of the intrinsic IFN-γ production by CD4+ T cells. This opinion was also explained the fact that CD4+ T cells were not reversible for the short-term reconstitution of exogenous p28 (Supporting Fig. 5). These CD4+ T cells still had an elevated IFN-γ production ability even after being transferred to hosts that had intact p28 in DCs (Fig. 5). Using the thymus transplantation and intrathymic injection approaches, we established the possibility that the functional development of CD4+ T cells in thymus requires the DC-derived p28 (Fig. 8). Although the IFN-γ regulation of CD4+ T cells in the periphery is not the focus of our current study, and other changes may occur in DCs of CD11c-p28f/f mice, no significant differences in DC development and function were observed, with the exception of p28 deficiency (Supporting Fig. 7A-E). Thus, for the first time, we have revealed a novel function of DC-derived p28 in CD4+ T cell regulation. This is a breakthrough discovery in CD4+ T cell biology, and also potentially explains the CD4+ T cell disorder in disease.

Similar to other IL-12 family cytokine subunits, p28 is also promiscuous and may combine with other molecules such as EBI3 or CLF to form a functional complex or be functional by itself. 10, 19, 33 Whether the indicated regulation on CD4+ T cells is caused by p28 alone or by a complex containing p28 remains to be determined.

In conclusion, by using p28 conditional knockout mice, our study reveals previously undefined and unique functions of p28 in the functional development of CD4+ T cells, resulting in a high response to both low and higher doses of ConA. Our conditional knockout mice may also provide a unique model to study the role of CD4+ T cells in the pathogenesis of liver inflammation and liver damage.

Acknowledgements

We thank Dr. Jie Du for providing CD4−/− mice, and we thank Dr. Richard Flavell for providing IFN−/− mice. We thank Dr. Mark Bartlam, Dr. Yisong Wan, and Dondavid S. Powell for manuscript revision. We thank Dr. Yu Zhang, Dr. Rong Jin, Xi Xu, and Chen Yin from Peking University Health Science Center (Beijing, China) for providing Rag2-EGFP mice, thymus experiment design, and technical support.

Ancillary