Apoptotic cells attenuate fulminant hepatitis by priming Kupffer cells to produce interleukin-10 through membrane-bound TGF-β


  • Minggang Zhang,

    1. National Key Laboratory of Medical Immunology & Institute of Immunology, Second Military Medical University, Shanghai, China
    2. Institute of Immunology, Zhejiang University School of Medicine, Hangzhou, China
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Sheng Xu,

    1. National Key Laboratory of Medical Immunology & Institute of Immunology, Second Military Medical University, Shanghai, China
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Yanmei Han,

    1. National Key Laboratory of Medical Immunology & Institute of Immunology, Second Military Medical University, Shanghai, China
    Search for more papers by this author
  • Xuetao Cao

    Corresponding author
    1. National Key Laboratory of Medical Immunology & Institute of Immunology, Second Military Medical University, Shanghai, China
    2. Institute of Immunology, Zhejiang University School of Medicine, Hangzhou, China
    • National Key Laboratory of Medical Immunology & Institute of Immunology, Second Military Medical University, 800 Xiangyin Road, Shanghai 200433, China
    Search for more papers by this author
    • Fax: +86 21 6538 2502

  • Potential conflict of interest: Nothing to report.


The liver, a unique tolerogenic organ, is regarded as the site to trap and destroy aging erythrocytes and activated T cells. However, to date, the mechanisms for why the liver is tolerogenic and whether liver Kupffer cells (KC) are critical phagocytes for apoptotic cells (AC) contributing to the liver immunosuppression remain unclear. Here we report that KC is the main phagocyte for AC in the liver. Contact of AC inhibits proinflammatory cytokine but enhances anti-inflammatory cytokine production of KC in response to lipopolysaccharide (LPS) stimulation. Membrane-bound transforming growth factor (TGF)-β on AC is responsible for the increased production of interleukin (IL)-10 in KC through extracellular signal-regulated kinase (ERK) activation via the Smad3 pathway. Importantly, KC-derived IL-10 is critical for AC infusion-mediated protection of endotoxin-induced fulminant hepatitis through suppression of tumor necrosis factor (TNF)-α and nitric oxide (NO) production from KC and consequently attenuation of KC-mediated cytolysis of hepatocytes. Conclusion: AC can be preferentially phagocytosed by KC in the liver, leading to attenuation of fulminant hepatitis through IL-10-mediated suppression of KC-derived inflammatory TNF-α and NO production. These findings demonstrate that priming of KC by AC may contribute to maintain liver immunosuppression, providing a new mechanistic explanation for how immune homeostasis is maintained in the liver. (HEPATOLOGY 2011.)

The liver is continuously exposed to lipopolysaccharide (LPS) and other pathogenic components from the gastrointestinal (GI) tract via the hepatic portal vein. LPS is a potent stimulator for immune responses, but normally these products from pathogenic microbes are cleared by the liver, leading to immune tolerance rather than significant immune response and inflammation in the liver.1 Hepatic tolerance is also reflected in immune tolerance toward oral antigens and liver grafts.2 Nonparenchymal liver cells, including liver sinusoidal endothelial cells, dendritic cells (DCs) and Kupffer cells (KCs), have been shown to be responsible for the creation of a local immunosuppressive microenvironment, permitting immune effector cells, such as T cells, to deliver tolerogenic signals, which contributes to the tolerogenic properties of the liver.3-6 However, the exact mechanisms for the hepatic immunosuppressive microenvironment remain to be fully understood.

The liver is the major “graveyard” of aging erythrocytes and neutrophils,7, 8 as well as a site for trapping and eliminating activated CD8+T and CD4+T cells that have completed their immunological functions and become apoptotic.9, 10 Phagocytic clearance of apoptotic cells (ACs) generally leads to immunosuppression.11 Contact with ACs during the resolution of inflammation or in remodeling tissue educates macrophages to adopt an immunoregulatory property.11-14 Ingestion of ACs by DCs and monocytes also stimulates their production of anti-inflammatory cytokines (e.g., transforming growth factor [TGF]-β and interleukin [IL]-10) and inhibits proinflammatory cytokines.11, 15 This interaction is poorly understood, but seems to require recognition of specific molecules on ACs, such as anionic lipid phosphatidylserine, by the phagocytic cells.12, 16

Strong immunological challenges will result in liver injury. On the basis of data from human clinical analysis and experimental animal models, nearly all innate immune cells are associated with diverse liver injury.17, 18 As for popularly used murine models, natural killer T cells (NKTs) mediate concanavalin A–induced hepatitis, and natural killer (NK) cells mediate polyinosinic:polycytidylic acid–induced hepatitis,17, 19 and KCs are reported to mediate endotoxin-induced fulminant hepatitis, which is associated with increased production of tumor necrosis factor (TNF)-α.19 However, the exact mechanisms of liver injury and normally how the liver balances the immune response and immunosuppression remain to be further determined.

Consistent with previous reports,11-15 we found administration of donor ACs can dramatically suppress the immune response and inflammation in a mouse model of endotoxin-induced hepatitis, which is mediated by IL-10 from KCs after priming with membrane-bound TGF-β on ACs. We further determined the molecular mechanisms for how IL-10 was induced and functioned. These results demonstrate that priming of KCs by ACs is an important component of the immunosuppressive microenvironment in the liver.


Ab, antibody; AC, apoptotic cells; ALT, alanine aminotransferase; ANOVA, analysis of variance; CFSE, 5,6-carboxyfluorescein diacetate succinimidyl ester; D-GalN, D-galactosamine; DT, diphtheria toxin; ELISA, enzyme-linked immunosorbent assay; ERK, extracellular signal-regulated kinase; FACS, fluorescence-activated cell sorting; FBS, fetal bovine serum; GdCl3, gadolinium chloride; HBSS, Hank's balanced salt solution; IL, interleukin; iNOS, nitric oxide synthase; i.p., intraperitoneal; i.v., intravenous; KC, Kupffer cells; LPS, lipopolysaccharide; mRNA, messenger RNA; NK, natural killer; NKT, natural killer T cell; NO, nitric oxide; p.v., portal vein; RPMI-1640, Roswell Park Memorial Institute 1640 medium; RT-PCR, reverse transcription polymerase chain reaction; SMT, S-methylisothiourea sulfate; TGF-β, transforming growth factor β; TNF-α, tumor necrosis factor α; Treg, regulatory T cell; UV, ultraviolet.

Materials and Methods


Adult male C57BL/6 (H-2b) and BALB/c (H-2d) mice were obtained from Sipper BK Experimental Animals Co. (Shanghai, China). C57BL/6 mice were used if there was no special mention. CD11c-DTR mice (C57BL/6 background) and IL-10-deficient mice (C57BL/6 background) were obtained from Jackson Laboratory (Bar Harbor, ME). Smad3-deficient mice were a gift from Dr. Xiao Yang (Beijing Academy of Medical Sciences, Beijing), and were established as described.20 Smad3−/− and Smad3+/+ (as wild-type control and origin of ACs for Smad3−/− KC priming) homozygous littermates were used (sexuality balanced), which were derived from the filial generation mice of Smad3+/− mice mating with Smad3+/− mice. All mice were housed in a SPF facility, and used at 6-8 weeks of age. All procedures were approved by the Scientific Investigation Board of the Second Military Medical University in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Sources of the antibodies and reagents are described in the Supporting Materials.

Cell Preparations.

Apoptotic splenocytes were obtained using an ultraviolet (UV)-B irradiation method as described,21 with a minor modification. Briefly, single-cell suspensions of mouse splenocytes were prepared and dead cells or debris were removed using 35% Ficoll. Five million cells were suspended in 5 mL Hank's balanced salt solution (HBSS) in a 10-cm-diameter Petri plate, irradiated using a 40-W UV-B (320 nm) at a distance of 40 cm for 10 minutes, and then cultured in Roswell Park Memorial Institute 1640 medium (RPMI-1640) (with 10% fetal bovine serum [FBS]) at 37°C in 5% CO2 for 4 hours. For apoptotic cell inoculation via vein or priming KCs in vitro, syngeneic apoptotic splenocytes were used, and 1.5×107 ACs were injected per mouse unless there is special mention. The purity of early ACs (propidium iodide-, annexin V+) was 90% to 95%. For phagocytosis assay, F4/80 splenocytes were sorted by Dako Moflo-XDP, and irradiated with UV-B after labeling with 5,6-carboxyfluorescein diacetate succinimidyl ester (CFSE). Splenic macrophages and thioglycolate-elicited peritoneal macrophages were prepared as described.22, 23 Hepatocytes and liver KCs were prepared from mice as described.17, 24 All cells were cultured for 1-3 days before being used in the experiments. For in vitro priming, KCs were incubated with ACs (1:10) for 4 hours, washed to get rid of the uncaptured ACs, and cultured for 3 days.

Fluorescence-Activated Cell Sorting Analysis.

Fc receptors were blocked using 2.4G2 and stained with fluorescent antibodies as described.17, 25 Fluorescence-Activated Cell Sorting (FACS) analysis was performed using BD LSR II with FACSDiva software (BD Biosciences). ERK phosphorylation was examined using a BD phosflow method as described.25, 26

Cell Depletion, Phagocytosis Inhibition, and Cell Adoptive Transfer.

KCs were depleted by intravenous (i.v.) injection of gadolinium chloride (GdCl3, 20 mg/kg body weight) in mice.27 For depletion of DCs, CD11c-DTR mice received intraperitoneal (i.p.) injection of diphtheria toxin (DT, 16 μg/kg body weight) at 24 hours prior to AC infusion.28 An anti-CD25 Ab (PC 61) was used to deplete regulatory T cells from mice at 24 hours before LPS/D-GalN injection.29 Cytochalasin B, a blocker of microfilament formation, was used to inhibit phagocytosis of macrophages and DCs.30 For adoptive cell transfer, mice were treated with GdCl3, and then received portal vein (p.v.) injection of AC-primed KCs (1 × 106/mouse) 24 hours later.

Detection of Cytokines and NO.

Cytokine concentration was measured using commercial enzyme-linked immunosorbent assay (ELISA) kits from R&D Systems (Minneapolis, MN). NO was assayed using a Griess reagent kit (Invitrogen, Eugene, OR).

Endotoxin-Induced Fulminant Hepatitis Model.

Fulminant hepatitis in mice was established by injection with LPS (5 μg/kg body weight) and D-GalN (400 mg/kg body weight) as described.27 Animal survival, liver histology, serum alanine aminotransferase (ALT) and cytokines were examined as described.17

Statistical Analysis.

Student t test was used to analyze statistical significance of differences for paired samples. One-way analysis of variance (ANOVA) was used to analyze differences between groups and a post hoc Bonferroni test to correct for multiple comparisons. Survival data were analyzed with Kaplan-Meier analysis. Statistical significance was set at P < 0.05.


KCs Are Critical for the Clearance of ACs in the Liver.

Apoptotic naive splenocytes i.v. infused via the tail vein were quickly captured and cleared (within 4 hours) by splenic macrophages (Fig. 1A). Apoptotic-activated splenocytes infused i.v. or apoptotic splenocytes infused p.v. (whether naive or activated) were mainly captured by hepatic KCs. To focus on the regulation of liver immunity by ACs, we chose p.v. infusion of apoptotic naive splenocytes for all subsequent in vivo experiments unless indicated otherwise.

Figure 1.

Kupffer cells are critical for phagocytic clearance of apoptotic cells in the liver. (A) CFSE+ F4/80+ cells in the spleen (upper panel) and liver (lower panel) of mice (n = 6) were quantified by FACS at indicated time after i.v. (tail vein) or p.v (portal vein) infusion of CFSE-labeled syngeneic AC (F4/80). (B) CFSE+ F4/80+ cells in the spleen (upper panel) and liver (lower panel) of mice (n =6) were quantified by FACS at indicated time after p.v. infusion of CFSE-labeled autologous AC (F4/80). AC was infused into syngeneic mice (control), syngeneic mice pretreated with GdCl3 (20 mg/kg, at −24 hours), syngeneic mice pretreated with cytochalasin B (Cyt B, 10 mg/kg, at −1 hour), or CD11c-DTR mice pretreated with DT (16 μg/kg, at −24 hours, CD11c deletion); n=6. (C) CFSE-labeled AC (F4/80) was incubated with F4/80+ macrophages from the spleen, peritoneal cavity, or KCs for 4 hours in vitro. Then the phagocytic index and CFSE MFI of macrophages were determined. Experiments in (A) and (B) were performed independently using C57BL/6 or BALB/c mice for 4 times with similar results, and data shown are from C57BL/6 mice; experiments in (C) were performed independently 3 times using C57BL/6 mice with similar results. Mϕ: macrophages.

Pretreatment with cytochalasin B significantly delayed the clearance of ACs in the liver (Fig. 1B). Depletion of KCs with GdCl3 also significantly delayed the clearance of ACs in the liver. In contrast, treatment of CD11c-DTR mice with DT did not change AC clearance in the liver. In addition, the AC capture capability of DC in vivo was not so potent compared with that of KCs in the liver or macrophages in the spleen (Supporting Fig. 1A,B). Furthermore, an in vitro experiment demonstrated that liver KCs had higher phagocytic capacity for ACs (Fig. 1C; Supporting Fig. 1C) but had similar pinocytic capacity in comparison to splenic and peritoneal macrophages (Supporting Fig. 1D). These results suggest that liver KCs but not DCs are the major phagocyte for AC capture and clearance in the liver.

Infusion of ACs Suppresses KC-Mediated Endotoxin-Induced Fulminant Hepatitis Nonspecifically.

LPS/D-GalN induced a fulminant hepatitis in mice with a high mortality within 12 hours. Infusion of donor ACs but not normal splenocytes (Supporting Fig. 2B,C) prior to LPS/D-GalN challenge, particularly via p.v. infusion, significantly attenuated liver injury (Fig. 2A,B; Supporting Fig. 3A,B). Moreover, there is cross-inhibition of hepatitis by ACs between C57BL/6 and BALB/c mice (Supporting Fig. 4), indicating that the suppression of hepatitis by ACs is nonspecific.

Figure 2.

Clearance of apoptotic cells suppresses endotoxin-induced hepatitis via Kupffer cells. (A) Histological analysis of liver tissue prior to LPS challenge (normal), at 6 hours after LPS/D-GalN injection alone (hepatitis) or with AC injection through i.v. (AC i.v.) or p.v. (AC p.v.). C57BL/6 mice were used in all above experiments. (B) C57BL/6 mice (n = 10) were pretreated with PBS or AC (1.5 × 107/mouse) via p.v. 3 days prior to the mice were injected with LPS/D-GalN, and their survival was observed. (C) Different groups of mice were observed for their survival after LPS/D-GalN treatment. Gdcl3: C57BL/6 mice (n = 8) received GdCl3; AC: ACs were injected at 3 days prior to LPS/D-GalN treatment; KC: p.v. injection of KCs at 1 day prior to LPS/D-GalN treatment; KC+AC: KCs and ACs were injected simultaneously at 3 days prior to LPS/D-GalN treatment; KC/AC: AC-primed KCs were injected at 1 day prior to LPS/D-GalN treatment. All data are representative of 5 independent experiments. CV: central vein.

Figure 3.

Priming with apoptotic cells affects the cytokine secretion of Kupffer cells. (A,B) KCs were cultured with ACs (1:10) for 4 hours, washed, and cultured for an additional 3 days. After stimulation with LPS (10 ng/mL) for 12 hours in fresh culture media, the supernatant was collected, and (A) inflammatory cytokines TNF-α, IL-6, and IL-1β and (B) anti-inflammatory cytokines IL-4, IL-10, and TGF-β were assayed. **P < 0.01. All data are representative of 3 independent experiments using C57BL/6 mice, and there are similar results in experiments using BALB/c mice (data not shown).

Figure 4.

IL-10 is responsible for reduced cytolysis of LPS-stimulated Kupffer cells to hepatocytes through suppression of TNF-α and NO production. (A) C57BL/6 KC was primed with syngeneic ACs for 4 hours, and then cultured alone for 3 days prior to stimulation with LPS for 12 hours. KCs were then cocultured with hepatocytes at the indicated E:T (KC:hepatocytes) ratio. The number of live hepatocytes was assayed using anti-F4/80-APC Ab, Annexin V-FITC, and 7-AAD. The cytolysis (%) = (1 − (7AADAnnexin V F4/80/F4/80)) × 100. (B) C57BL/6 wild-type (WT) KCs were primed with syngeneic ACs as described in A, and then stimulated with LPS. The syngeneic hepatocytes were mixed with KC cells at an E:T of 1:10. PD98059 (25μM) was added at 1 hour before AC priming. Anti-IL-4 Ab, anti-TGF-β Ab, anti-IL-10 Ab, anti-TNF-α Ab, S-methylisothiourea sulfate (SMT) or anti-TNF-α Ab plus SMT were added immediately prior to LPS stimulation. **P < 0.01. (C) IL-10-deficient KCs were primed with ACs as described in A, and then stimulated with LPS. The hepatocytes were mixed with KCs at an E:T of 1:10. The cytolysis of hepatocytes is shown. (D) C57BL/6 KCs were primed with syngeneic ACs for 4 hours and cultured for 3 days, then stimulated with LPS for 12 hours in the presence or absence of anti-IL-10 Ab. TNF-α and NO production by KCs was assayed. **P < 0.01. Data are representatives of 3 experiments with similar results.

The timing of AC infusion was critical, with maximum protective effects at 3-7 days and much less effective within 3 days, prior to the LPS/D-GalN challenge (Supporting Fig. 3C,D). So, we selected to infuse ACs at 3 days before the LPS/D-GalN challenge for all relevant subsequent in vivo experiments.

Maximum protective effects of the AC infusion were observed at doses of 1 × 107 to 3 × 107 ACs per mouse (p.v.; Supporting Fig. 3E,F). Somewhat surprisingly, a higher dose of the AC infusion at 2 × 108 per mouse had no protective effects. So, p.v. infusion of 1.5 × 107 ACs/mouse was used in all subsequent in vivo experiments.

Depletion of KCs by GdCl3 completely prevented LPS/D-GalN-induced hepatitis (Fig. 2C; Supporting Fig. 2A). Adoptive transfer of KCs, but not AC-primed KCs, restored the hepatic susceptibility to LPS/D-GalN. However, depletion of either regulatory T cells (Supporting Fig. 5A,B) or DCs (Supporting Fig. 5C,D) could not eliminate these protective effects, indicating that regulatory T cells and DCs are not involved in the protective effect of AC infusion. In addition, at 3 days after AC inoculation, no change of T helper 1/T helper 2 (Th1/Th2) cells was found (Supporting Fig. 6), suggesting that Th1/Th2 balance may not involved in the protection by ACs. These results suggest that KCs are critical to mediate LPS/D-GalN-induced fulminant hepatitis, and AC infusion can suppress KC-mediated LPS/D-GalN-induced fulminant hepatitis.

Figure 5.

Kupffer cell–derived IL-10 is responsible for the suppression of Kupffer cell–mediated fulminant hepatitis. (A) Serum IL-10 level was assayed in C57BL/6 mice (n = 6) at the indicated time after LPS/D-GalN injection. AC: AC (1.5 × 107/mouse, p.v.) was injected at 3 days prior to LPS/D-GalN injection, GdCl3: GdCl3 was injected at 24 hours before LPS/D-GalN injection. GdCl3+AC: AC was injected p.v. at 24 hours after GdCl3 injection, and 3 days later, LPS/D-GalN was injected. (B) KCs derived from wild-type (WT) mice or IL-10-deficient (IL-10−/−) mice were p.v. injected into WT C57BL/6 mice (n = 10) at 24 hours after GdCl3 infusion. One day later, these mice were injected with LPS/D-GalN. AC: mice received LPS/D-GalN at 3 days after AC injection; PBS: mice received LPS/D-GalN at 3 days after PBS injection; KC/AC: AC-primed KCs. The survival of the mice receiving the different treatments was observed. (C) Serum ALT of the mice in B at 6 hours after LPS/D-GalN injection. Normal: mice receiving no treatment. **P < 0.01. (D) IL-10–deficient (IL-10−/−) mice (n = 10) received LPS/D-GalN at 3 days after AC injection. The survival of the mice was observed. (E) Serum ALT of mice in D at 6 hours after LPS/D-GalN injection. Normal: mice receiving no treatment. Experiments were performed 3 (A) or 5 (B-E) times with similar results.

Figure 6.

Membrane-bound TGF-β on apoptotic cells is responsible for increased IL-10 production by apoptotic cell–primed Kupffer cells. (A) Supernatant of ACs cultured for 6 hours (Supernatant) was used to culture KCs. Transwell was used to separate ACs (upper chamber) from KCs (lower chamber). Cytochalasin B (Cyt B) pretreated KCs were cocultured with AC. Four hours later, the KCs were washed and cultured with fresh culture medium for an additional 3 days prior to LPS stimulation (10 ng/mL for 12 hours). IL-10 in the supernatant was measured with ELISA. Both KC and AC were derived from C57BL/6 mice. **P < 0.01. (B) In the coculture system of KCs and ACs, transwell was used to separate the upper ACs and the lower KCs. Anti-IL-10, anti-TGF-β, or isotype antibody was added to the KC culture just before AC addition. And recombinant murine TGF-β was used to replace ACs (+TGF-β). Four hours later, the KCs were washed and cultured for 3 days with fresh culture medium. IL-10 in supernatant was assayed after LPS stimulation for 12 hours. Both KCs and ACs were derived from C57BL/6 mice. **P < 0.01. (C) Smad3+/+ wild-type (WT) or Smad3-deficient (Smad3−/−) KCs were cocultured with ACs derived from homozygous littermates Smad3+/+ WT mice for 4 hours, then washed to get rid of the uncaptured ACs, cultured for 3 days, and stimulated with LPS for 12 hours. IL-10 concentration in the supernatant was measured using ELISA. **P < 0.01. (D) Membrane-bound TGF-β on ACs was determined by FACS at the indicated time after UV-B irradiation. The membrane-bound TGF-β–positive cell percentage is shown. Isotype: isotype antibody staining of cells 2 hours after UV-B irradiation. Data are representative of 3 independent experiments.

Priming by ACs Inhibits Proinflammatory Cytokines but Enhances Anti-Inflammatory Cytokine Production of KCs in Response to LPS Stimulation.

When cultured alone, KCs appeared to be rounded and relatively nonadherent to the culture wall. Three days after AC priming, KCs became firmly adherent and ellipsoid in shape (Supporting Fig. 7A). Upon overnight culture with 10 ng/mL LPS, KCs became more tightly adherent and the cell skirt stretched out. Surface expression of costimulatory molecules (CD40, CD80, and CD86), CD14, and B7-H1 on AC-primed KCs remained comparable to the controls (Supporting Fig. 7B). AC priming alone did not affect KC production of cytokines. However, LPS-induced cytokines were significantly altered by AC-primed KCs: proinflammatory cytokines (TNF-α, IL-6, IL-1β) were significantly reduced (Fig. 3A); anti-inflammatory cytokines (IL-4, IL-10, and TGF-β) were significantly enhanced (Fig. 3B).

Figure 7.

ERK pathway activated by membrane-bound TGF-β on apoptotic cells is responsible for increased IL-10 production in Kupffer cells. (A) C57BL/6 KCs were primed by syngeneic ACs for 4 hours, then uncaptured ACs were removed, and cultured for 3 days, then stimulated by LPS for 12 hours. IL-10 expression was examined with RT-PCR analysis. (B) C57BL/6 KC was stimulated with syngeneic ACs and LPS as described above. At 1 hour before adding ACs, PD98059, a specific ERK inhibitor, was added. Three days later, KC was stimulated by LPS for 12 hours. IL-10 in the supernatant was determined using ELISA. *P < 0.5, **P < 0.01. (C) Thirty minutes after adding ACs or LPS, KCs were collected for analysis of pERK using a Phosflow method; pERK: fluorescence intensity (MFI) of all F4/80–positive cells is shown. Culture medium was used as a negative control; LPS was used as a positive control. Transwell was used to separate the upper ACs and the lower KCs. Anti-TGF-β, the neutralizing anti-TGF-β antibody, was added before AC priming. TGF-β: recombinant murine TGF-β was used instead of ACs. Smad3−/−: Smad3-deficient KCs were primed by ACs. All KCs except for Smad3−/− mice were derived from C57BL/6 mice stimulated with syngeneic ACs (Smad3−/− KCs were stimulated with ACs from Smad3+/+ homozygous littermates). All data are representative of 3 independent experiments. pERK: phosphorylated ERK.

Priming by ACs Reduces the Cytotoxicity of LPS-Triggered KCs to Hepatocytes by IL-10-Mediated Suppression of TNF-α and NO Production.

LPS-stimulated KCs exhibited potent cytotoxicity against hepatocytes. Coculture with KCs that were not primed with ACs were potent cytotoxic to hepatocytes, decreasing the number of live hepatocytes significantly. This cytotoxic effect of KCs was significantly attenuated by AC priming or an anti-TNF-α antibody (Fig. 4A,B). In addition, in vitro use of neutralizing anti-IL-10 antibodies (Abs), but not anti-IL-4 or anti-TGF-β Abs, could eliminate the inhibition of hepatocytes cytotoxicity by ACs (Fig. 4B). And if we used IL-10-deficient KCs instead of wild-type KCs as above, AC priming could not inhibit the cytotoxicity of KCs and prevent decrease of live hepatocytes number (Fig. 4C). S-methylisothiourea sulfate (SMT), a specific inducible nitric oxide synthase (iNOS) inhibitor, also dramatically reduced the death of hepatocytes induced by KCs. An anti-TNF-α antibody significantly enhanced the inhibitory effect of SMT on KC-mediated death of hepatocytes. Furthermore, neutralizing anti-IL-10 Abs reversed the inhibition of TNF-α and NO production of KCs by AC priming (Fig. 4D), consistent with a previous report that IL-10 can target TNF-α messenger RNA (mRNA) to inhibit its translation.31 These data suggest that IL-10 derived from AC-primed KCs could inhibit the KC-mediated cytotoxicity of hepatocytes through inhibition of TNF-α and NO production.

KC–Derived IL-10 Is Responsible for the Suppression of KC–Mediated Endotoxin-Induced Fulminant Hepatitis by ACs.

Considering the importance of IL-10 in immune suppression, serum IL-10 level in AC-preinjected mice was assayed, showing significant increase after LPS/D-GalN injection (Fig. 5A). In normal mice without AC injection, LPS/D-GalN stimulation could increase the IL-10 production in the serum at a very low level, with increase at 4 hours but decrease at 8 hours after the LPS/D-GalN injection (Supporting Fig. 8). Once KCs were made dysfunctional with GdCl3 prior to AC infusion, the IL-10 response was significantly inhibited (Fig. 5A), suggesting KCs were the major producer of IL-10 after AC infusion. Furthermore, infusion of IL-10-deficient KCs, either naive or AC-primed, into mice already depleted of KCs with GdCl3 did not protect the liver from LPS challenge (Fig. 5B,C). Also, infusion of ACs to IL-10-deficient mice did not protect the liver from LPS-induced damage (Fig. 5D,E). Thus, IL-10 derived from KCs is critical for the suppression of endotoxin-induced hepatitis by AC infusion.

Membrane-Bound TGF-β on ACs Is Responsible for the Increased IL-10 Production of KCs.

Consistent with a previous report,32 cultured-alone ACs released IL-10 and TGF-β upon UV-B irradiation (Supporting Fig. 9). However, adding supernatant from culture-alone ACs did not induce IL-10 production of KC culture, suggesting soluble factor(s) from ACs are not responsible for the increased IL-10 production of KCs. In the KC and AC coculture system, using transwell (0.4μM pore size) to separate KCs from ACs dramatically reduced IL-10 production (Fig. 6A,B). Blockade of phagocytosis by cytochalasin B could not prevent IL-10 production of KC to AC priming, suggesting that cell-cell contact between KCs and ACs, but not phagocytosis, is critical for IL-10 induction.

Neutralizing anti-TGF-β1 but not anti-IL-10 Ab blocked IL-10 production of KC to AC priming (Fig. 6B). However, the recombinant murine TGF-β1 did not increase IL-10 production as ACs did. Smad3-deficient mice are deficient in TGF-β signal transduction,25, 33 and hence were used to further examine the role of TGF-β. AC priming did not increase IL-10 production in KCs from Smad3-deficient mice (Fig. 6C). Consistently, when we transferred Smad3−/− KCs primed with ACs to GdCl3-pretreated mice, the protective effect of ACs on hepatitis disappeared (Supporting Fig. 10A,B). Intriguingly, FACS analysis revealed a significant increase of membrane-bound TGF-β1 on ACs (Fig. 6D), and up to 56.9% cells expressed membrane-bound TGF-β1 at 4 hours after UV-B irradiation. So, it is membrane-bound TGF-β on ACs that induces IL-10 production from KCs.

ERK Pathway Activated by Membrane-Bound TGF-β on ACs Is Responsible for the Increased IL-10 Production in KCs.

Reverse transcription polymerase chain reaction (RT-PCR) assays showed that, after AC priming, IL-10 mRNA of KCs increased significantly in response to LPS (Fig. 7A). We observed that IL-10 production of KCs in response to AC priming could be significantly reduced when the ERK pathway was suppressed by PD98059, a specific inhibitor of ERK (Fig. 7B), but inhibition of the JNK or p38 pathway could not influence the IL-10 production (data not shown). Furthermore, FACS analysis revealed that contact with ACs increased phosphorylation of ERK in KCs (Fig. 7C; Supporting Fig. 11). A neutralizing antibody against TGF-β1 eliminated the ERK activation of KCs in response to AC priming. However, ERK in wild-type KCs was not activated by recombinant murine TGF-β, and not in KCs from Smad3-deficient mice in response to AC priming. Thus, membrane-bound TGF-β on ACs activates the ERK pathway in KCs, leading to IL-10 production.


Hepatic phagocytes, including hepatocytes, KCs, liver sinusoidal endothelial cells, and DCs, could engulf ACs in vivo.34 In this study, we have demonstrated: 1) KCs preferentially capture ACs in the liver and; 2) recognition, cell contact, and clearance of ACs by KCs suppresses the induction of endotoxin-induced hepatitis. More KCs would be primed if ACs were administered via p.v., leading to more IL-10 production and more potent protection against hepatitis than that of other injection routes. Intriguingly, the best protection against LPS-induced liver injury occurred when ACs were administered at 3-7 days prior to the LPS challenge, suggesting a need for at least 3 days of AC priming to induce enough IL-10 from KCs to prevent LPS-induced liver injury. In addition, infusion of 1 × 107–3 × 107 ACs gain the best protection against hepatitis, suggesting redundant ACs beyond the clearance ability of KCs may become necrotic in the liver. Accordingly as described,35, 36 dysfunction of AC clearance may result in the pathogenesis of autoimmune diseases, such as systemic lupus erythematosus or diabetes.

Canbay et al.37 reported that KCs preprimed by apoptotic bodies could induce hepatocyte death and inflammation. However, the experimental system in Canbay et al.'s study was different from ours. They used apoptotic bodies from hepatocytes 48 hours after UV irradiation, and in an animal model after the bile duct ligation, strong bile chemical stimulation may induce a large amount of cells to be apoptotic and then become late apoptotic (necrotic); in contrast, we used ACs cultured just 4 hours in vitro after UV-B irradiation. After 24 hours, once UV-B irradiation (for mouse splenocytes) has been applied, most cells (>95%) will actually be at the late apoptotic stage (propidium iodide [PI]+ Annexin V+/−) (data not shown). Also, after 24 hours once UV-B irradiation has been applied, membrane-bound TGF-β will almost disappear on ACs (Fig. 6). In addition, necrotic cells can release DNA, heat shock protein (HSP), uritic acid, and other inflammatory-stimulating factors to contribute to inflammation development.38 Appropriate amounts of AC infusion will be cleared very quickly in vivo (<4 hours; Fig. 1A), possibly there are few ACs becoming necrotic before being phagocytosed normally. So, ACs or apoptotic bodies prepared from different systems at different time points may induce different effects, proinflammatory or anti-inflammatory, once infused in vivo.

Clearance of ACs by phagocytes typically inhibits the release of proinflammatory cytokines and increases anti-inflammatory cytokines.15, 39 IL-10 is an important anti-inflammatory cytokine,40 and plays a central role in the attenuation of inflammation in a variety of diseases including hepatitis.41 Increased IL-10 production along with decreased proinflammatory cytokines has been noted in phagocytes treated with ACs and LPS.14, 42 ACs increase generation of IL-10 by macrophages in a cell-to-cell contact–dependent, but not phagocytosis-dependent manner.14 A previous study demonstrated that under physiological conditions, KCs generate more IL-10 in response to LPS than other macrophages,43 suggesting a preferential interaction of KCs for ACs, such as apoptotic neutrophils or T cells.8-10 However, the number of ACs is much lower under physiological conditions than were used in this experiment, thus only a few KCs were primed and produced only a small amount of IL-10, with moderate immunosuppression in physiological conditions. Even if after p.v. AC infusion, possibly not all KCs were primed after engulfing ACs, but IL-10 produced from AC-primed KCs may continually inhibit non-AC-primed KCs, leading to suppression of immune responses in the liver.

Using apoptotic splenocytes, Sun et al.21 successfully induced donor-specific tolerogenicity, significantly prolonging allograft survival. However, the reports, including our data (Supporting Fig. 4), show that AC infusion could induce nonspecific immune inhibition by production of anti-inflammatory cytokines.11, 12, 32 The inconsistency may be due to the different times spent observing after AC inoculation. Three days after AC injection, the immune inhibition is nonspecific, but donor-specific immune tolerance may occur 7-14 days later. And 7-14 days is consistent with the time of regulatory T cell (Treg) induction. Treg induction is not significant at 3 days, but started to increase at 7 days, and reached a peak at 14 days after the AC injection (our unpublished data). Therefore, we suggest that phagocytes may be primed to become suppressive in the early stage after AC injection, releasing anti-inflammatory cytokines and forming a nonspecific inhibitory microenvironment. However, 7 days later, donor-specific tolerogenicity may be established following the Treg induction.

Previous studies have indicated that phosphatidylserine, an anionic aminophospholipid on the surface of ACs, plays an important role in the recognition and clearance of ACs by macrophages.12 Interaction of macrophages with phosphatidylserine results in the suppression of macrophages in response to LPS.12 Also, ACs themselves could release soluble IL-10 and TGF-β after UV-B irradiation, and it is reported that these cytokines from ACs contribute to an immunosuppressive milieu,32 but we found these soluble factors released from ACs are not involved in the priming of KCs to be immunosuppressive. As for TGF-β, it can work in a membrane-bound manner as well as in a soluble manner.25, 33, 44 We showed that membrane-bound TGF-β is highly expressed on ACs. Production of IL-10 by KCs was reversed by separating ACs from KCs using transwell in vitro, and blockade of TGF-β using neutralizing anti-TGF-β antibody in vitro, suggesting a critical role of membrane-bound TGF-β on ACs. Membrane-bound TGF-β has been reported to suppress immune function via many other mechanisms, including inhibition of cytotoxic T lymphocytes (CTLs) responses, induction of CD8+ T cell anergy, and CD4+ Tr1 cell responses, and down-regulation of NKG2D in NK cells.25, 33, 44, 45 It is plausible that membrane-bound TGF-β on ACs can also directly inhibit immune responses via these processes. Nonetheless, in the current study we discovered that membrane-bound TGF-β on ACs could lead to hepatic immune hyporesponsiveness via induction of IL-10 from KCs.

Interestingly, neutralization of TGF-β could suppress IL-10 production, but soluble recombinant TGF-β did not induce IL-10 in KCs. Previous reports showed the similar functional difference between membrane-bound TGF-β and recombinant TGF-β25, 33, 44, 45; however, the mechanisms remain to be clarified. One hypothesis is that the membrane-bound TGF will have a kind of group effect, and a very high dose of soluble TGF may own the effect of membrane-bound TGF. Another possibility is that membrane-bound TGF-β may sustain some signaling, which is required for activation of certain signaling, leading to IL-10 increase in KCs that cannot be achieved by soluble TGF-β. And the third possibility is that the synergistic effect of membrane-bound TGF-β with other adherent molecules on the cell surface may be very important for its special function versus soluble TGF-β.

Smad signaling is critical for TGF-β,46 and cells from Smad3-deficient mice lack TGF-β signaling.33 KCs from Smad3-deficient mice showed impaired IL-10 induction by AC priming. There was crosstalk between Smad and ERK.46 Our experiments with the specific ERK inhibitor PD98059 showed that ERK is required for increased IL-10 secretion from KCs after AC priming. In contrast, IL-10 secretion by KCs after AC priming was not affected by SP600125 (JNK-specific inhibitor) or SB203580 (p38-specific inhibitor),17 suggesting that the JNK or p38 pathways are not involved in this process. We also showed that membrane-bound TGF-β on ACs could activate ERK, and both Smad3 and ERK are required for increased IL-10 production in KCs after AC priming, suggesting membrane-bound TGF-β may regulate IL-10 production in KCs through the ERK/Smad3 pathway.

In summary, results from our current study indicate that membrane-bound TGF-β on ACs stimulates IL-10 production by KCs via activation of ERK. Then, IL-10 inhibits TNF-α and NO generation from KCs, and by doing so inhibits immune/inflammatory responses of the liver to LPS/D-GalN, preventing KC-mediated endotoxin-induced fulminant hepatitis. We provide a new pathway for suppressing liver injury by AC.


We thank Ms. Xiaoting Zuo and Ms. Jianqiu Long for technical assistance. We also thank Dr. Zheng Fu from the Statistics Department of Second Military Medical University for his helpful discussion.