Potential conflict of interest: Nothing to report.
We have documented the key role of toll-like receptor 4 (TLR4) activation and its signaling pathway mediated by interferon (IFN) regulatory factor 3, in the induction of inflammation leading to the hepatocellular damage during liver ischemia/reperfusion injury (IRI). Because type I IFN is the major downstream activation product of that pathway, we studied its role in comparison with IFN-γ. Groups of type I (IFNAR), type II (IFNGR) IFN receptor–deficient mice, along with wild-type (WT) controls were subjected to partial liver warm ischemia (90 minutes) followed by reperfusion (1-6 hours). Interestingly, IFNAR knockout (KO) but not IFNGR KO mice were protected from IR-induced liver damage, as evidenced by decreased serum alanine aminotransferase and preservation of tissue architecture. IR-triggered intrahepatic pro-inflammatory response, assessed by tumor necrosis factor (TNF-α), interleukin 6 (IL-6), and chemokine (C-X-C motif) ligand 10 (CXCL-10) expression, was diminished selectively in IFNAR KO mice. Consistent with these findings, our in vitro cell culture studies have shown that: (1) although hepatocytes alone failed to respond to lipopolysaccharide (LPS), when co-cultured with macrophages they did respond to LPS via macrophage-derived IFN-β; (2) macrophages required type I IFN to sustain CXCL10 production in response to LPS. This study documents that type I, but not type II, IFN pathway is required for IR-triggered liver inflammation/damage. Type I IFN mediates potential synergy between nonparenchyma and parenchyma cells in response to TLR4 activation. (HEPATOLOGY 2007.)
Liver ischemia/reperfusion injury (IRI) occurs in multiple clinical settings including surgical interventions, trauma, and transplantation.1–3 The mechanisms underlying liver IRI are complex but are known to involve interactions between both nonparenchyma cells, such as Kupffer cells (KC), and parenchyma cells, such as hepatocytes. Local leukocyte sequestration and activation (neutrophils, macrophages, and T cells) leads to the formation of reactive oxygen species, secretion of pro-inflammatory cytokines/chemokines, complement activation, and vascular cell adhesion molecule activation. Because IRI develops in the absence of exogenous antigens, it has been considered as an innate immune-mediated pro-inflammatory response.
It was demonstrated that the mammalian sentinel toll-like-receptor (TLR) system plays a critical role in the development of IRI.4–7 Indeed, TLR4 activation proved essential in the induction of inflammation in a warm liver IRI mouse model. In the absence of TLR4 but not TLR2 signaling, livers were protected from IRI, and intrahepatic induction of tumor necrosis factor (TNF-α) and interleukin 1 (IL-1β) was abolished.7 It was also shown, by using bone marrow chimeras, that TLR4 expression on hematopoietic rather than parenchyma cells was instrumental in promoting liver IR-mediated damage.8
TLR4 activation triggers 2 distinct signaling pathways leading to the induction of different immune-related genes. The MyD88-dependent pathway causes early phase nuclear factor kappa B activation, resulting in the production of pro-inflammatory cytokines; the MyD88-independent pathway activates interferon-regulatory factor 3 (IRF3) and causes the late-phase nuclear factor kappa B activation, both of which lead to the production of interferon (IFN-β) and IFN-inducible genes.9 We have shown that an IRF3, but not MyD88, dependent pathway is essential in the development of liver IRI.7 We have also found that CXCR3 ligands, including IFN-inducible CXCL9, CXCL10, and CXCL11, were up-regulated early during reperfusion, whereas blockade of their CXCR3 receptor ameliorated liver damage.10 Our recent study has documented that anti-oxidant heme oxygenase-1 xerts cytoprotective and anti-inflammatory functions by down-regulating STAT1 via the type 1 IFN pathway, which in turn decreases CXCL-10 production.11 The latter plays a key role in the induction of pro-inflammatory response that culminates in IR-induced hepatocellular damage (Zhai et al., unpublished observations.)
Type I IFN α/β are pleiotropic cytokines, produced mainly by macrophages/fibroblasyts. They bind to a common cell surface receptor complex known as the type I IFN receptor (IFNAR), consisting of subunits IFNAR1 and IFNAR2.2 Type II IFN-γ, elaborated primarily by T/natural killer cells and macrophages, binds to surface receptor IFNGR (deficient in type II IFN), made out of IFNGR1 and IFNGR2 subunits. Both IFNs exert antiviral, antiproliferative, and immune regulatory functions via Janus kinases and signal transducers and activators of transcription signaling. Because type I IFN is the major product of IRF3 pathway downstream of TLR4, the obvious question in dissecting molecular mechanisms of TLR4 function is the role of type I IFN in the pathogenesis of liver IRI. Although type II (IFN-γ) has been implicated in liver12, 13 and kidney14 IRI, and often used as a parameter in assessing local inflammation response, its role in IR is not well defined. To resolve these key issues, we first used IFNAR-deficient versus IFNGR-deficient mice and determined that type I IFN receptor KO mice proved resistant to liver IRI. Then, we studied cell–cell interactions in vitro and used well-defined liver cell substitutes, macrophage/hepatoma cell co-culture systems, in which immune mechanisms can be better controlled and more accurately defined. In agreement with in vivo findings, we identified type I IFN pathway as the key mediator responsible for potential synergy between the 2 cell types in response to TLR4 activation.
FN, interferon; IFNAR, type I interferon; IFNGR, type II IFN; IL, interleukin; IRF3, interferon-regulatory factor 3; IRI, ischemia/reperfusion injury; KC, Kupffer cells; LPS, lipopolysaccharide; TLR, toll-like receptor; TNF, tumor necrosis factor; WT, wild-type.
Materials and Methods
Male wild-type (WT; C57BL/6) mice (8-12 weeks old) were used (The Jackson Laboratory, Bar Harbor, ME). Mice deficient in type I IFN receptor (IFNAR) or type II IFN receptor (IFNGR) (C57BL/6 background) were obtained from Dr. Genhong Cheng (UCLA). Animals were housed in the University of California Los Angeles animal facility under specific pathogen-free conditions and received human care according to the criteria outlined in the “Guide for the Care and Use of Laboratory Animals” prepared by the National Academy of Sciences and published (NIH publication 86-23 revised 1985).
Mouse Warm Hepatic IRI Model.
We have developed a warm hepatic IRI model in mice.7 Briefly, mice were injected with heparin (100 g/kg), and an atraumatic clip was used to interrupt the arterial and portal venous blood supply to the cephalad lobes. After 90 minutes of partial liver warm ischemia, the clip was removed, initiating hepatic reperfusion. Mice were sacrificed after various times of reperfusion. Liver, spleen, and peripheral blood samples were collected. Serum alanine aminotransferase levels, an indicator of hepatocellular injury, were measured using an auto analyzer (ANTECH Diagnostics, Los Angeles, CA). Liver specimens were fixed in 10% buffered formalin and embedded in paraffin. Liver sections (4 μm) were stained with hematoxylin-eosin, and then analyzed blindly. Sham WT controls underwent the same procedure but without vascular occlusion.
Murine bone marrow macrophages were differentiated from marrow of 6-week-old to 10-week-old C57B/6 mice (WT, IFNAR KO, or IFNGR KO) by culturing in 1 × Dulbecco's modified Eagle's medium, 10% fetal bovine serum, 1% penicillin/streptomycin, and 30% L929 conditioned medium for 6 days. The cell purity was assayed to be 94% to 99% CD11b+. In addition, we used mouse macrophage and hepatoma lines (TIB-71, CRL-1830 ATCC, Manassas, VA). These were maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum. Cells were treated with lipopolysaccharide (LPS) (1 ng-10 μg/mL, Sigma, St. Louis, MO), or IFN-β (10-1000 U/mL, R&D Systems, Minneapolis, MN) for 2 to 24 hours. Treatment did not affect cell viability (>95%).
For IFN-β neutralization in vitro, rabbit polyclonal anti-rat IFN-β antibody (R&D Systems) was used. To test its efficacy, hepatoma cells were seeded 24 hours earlier at 400,000 cells/well. IFN-β1 antibody (or rabbit immunoglobulin G) was added first into cell cultures (150 U/mL), followed by rIFN-β (100 U/mL). Cells were incubated for 4 hours before harvesting for reverse transcription polymerase chain reaction (RT-PCR) analysis. Cell co-cultures were performed with transwells (0.4 μm pore size; Costar, Corning Inc., Corning, NY) in which macrophages were seeded in transwells, and hepatoma cells in 6-well plates. Anti-IFN-β antibody (150 U/mL) was added first in the hepatocyte-containing wells, followed by LPS (100 ng/mL) added in the macrophage-containing transwells. Co-cultures were incubated for 4 hours before harvesting hepatoma cells for gene expression.
RNA (2.5 g) was reverse-transcribed into complementary DNA using SuperScriptTM III First-Strand Synthesis System (Invitrogen, Carlsbad, CA). Quantitative PCR was performed using the DNA Engine with Chromo 4 Detector (MJ Research, Waltham, MA). In a final reaction volume of 25 μL, the following were added: 1 × SuperMix (Platinum SYBR Green qPCR Kit, Invitrogen, Carlsbad, CA), complementary DNA, and 0.5 mM of each primer. Amplification conditions were: 50°C (2 minutes), 95°C (5 minutes), followed by 50 cycles of 95°C (15 seconds), 60°C (30 seconds). Primers used to amplify a specific mouse gene fragments have been published.7, 15
Results are shown as mean ± standard deviation. Statistical analyses were performed using unpaired Student t test with P < 0.05 (2-tailed) considered as significant.
Liver IR Induces IFN-β But Not IFN-γ.
To analyze IFN gene induction profile during liver IR, the ischemic lobes were harvested serially after reperfusion. Sham-operated livers served as controls. Because the IFN-β gene does not have an intron, RNA with or without RT were subjected to quantitative RT-PCR (n = 3-4/group). The messenger RNA levels were calculated by subtracting levels derived from RNA without RT from that with RT. As shown in Fig. 1 (lower panel), there was an early 5-fold induction of IFN-β at time 0, that is, right after 90 minutes of warm ischemia, but before the onset of reperfusion; its full induction, approximately 100-fold higher as compared with controls, was detected at 6 hours after reperfusion. In contrast, IFN-γ gene induction remained undetectable throughout the observation period (Fig. 1; upper panel).
Type I, But Not Type II, IFN Pathway Is Required in Liver IRI.
To directly test the functional significance of type I versus type II IFN pathway in liver IRI, mice deficient of IFNAR or IFNGR were contrasted with WT counterparts (n = 4-9/group). Indeed, IFNAR KO mice were fully protected from IRI, as shown by both reduced serum alanine aminotransferase levels (254.8 ± 35.3, versus 3355 ± 415.5, P < 0.0005, Fig. 2A) and well-preserved liver architecture by histology (Fig. 2B). In contrast, INFGR KO mice remained susceptible to liver IRI and suffered even more severe hepatocellular damage than WT, as shown by both serum alanine aminotransferase levels (7988 ± 2634; P < 0.04 versus WT) and histology. Consistent with the functional data, intrahepatic pro-inflammatory gene induction was reduced in IFNAR KO but not IFNGR KO hosts, as shown by qRT-PCR analysis of TNF-α, IL-6, and CXCL10 genes (Fig. 2C). Interestingly, IL-1β gene induction was comparable in the 3 recipient groups. Thus, type I but not type II IFN pathway was instrumental for the development of liver inflammation and hepatocellular injury attributable to ischemia followed by reperfusion.
Macrophages Require Type I IFN to Sustain the Response to TLR4 Stimulation.
Next we studied the response of mouse macrophages to TLR4 stimulation in vitro. Bone marrow macrophages from groups of WT, type I, and type II IFN receptor KO mice were stimulated with LPS for 2 or 6 hours before harvesting and qRT-PCR analysis. As shown in Fig. 3, LPS-induced TNF-α induction was comparable in the 3 cell types at 2 hours. However, TNF-α levels decreased in IFNAR KO cells at 6 hours (2 hours versus 6 hours, P < 0.03). In contrast, IFNGR KO and WT cells had sustained TNF-α production. The levels and kinetics of IL-1β were similar in all 3 groups, whereas those of IL-6, although comparable early, did decrease by 6 hours in KO cells, as compared with WT. Most interestingly, however, IFNAR KO cells had reduced CXCL10 levels at 2 hours, which were further diminished at 6 hours. In marked contrast, no difference in CXCL10 production was found between IFNGR KO and WT cells. These results indicate that type I but not type II IFN pathway was required for macrophages to sustain pro-inflammatory response to TLR4 stimulation.
Macrophages But Not Hepatoma Cells Respond to LPS Stimulation.
We then turned to well-defined in vitro macrophage and hepatoma line systems to address some key mechanistic questions on the role of type I IFN in TLR4 activation. First, to determine LPS-induced pro-inflammatory cytokine/chemokine programs, cells were cultured with LPS (10 ng/mL), followed by qRT-PCR assessment at 4 hours. As shown in Fig. 4, addition of LPS readily up-regulated pro-inflammatory macrophage genes. Indeed, TNF-α/IL-6 expression increased 100-fold to 300-fold, as compared with unstimulated controls (Fig. 4A); macrophage/neutrophil-targeted CXCL1/CCL2 increased 20-fold to 30-fold (Fig. 4B), whereas T cell–targeted CXCL9, CXCL10, and CXCL11 varied (Fig. 4C). Thus, CXCL10 was up-regulated by more than 200-fold, CXCL11 approximately 40-fold, and CXCL9 showed minimal changes. TNF-α/CXCL10 protein production from LPS-stimulated culture supernatants was confirmed by enzyme-linked immunosorbent assay (not shown), and correlated with PCR data. In contrast, LPS (up to 10 μg/mL) failed to stimulate hepatoma cells to produce any of the cytokines at 4 hours (Fig. 4D) or 24 hours (not shown). Thus, macrophages but not hepatoma cells responded directly to LPS-induced TLR4 stimulation by producing pro-inflammatory cytokine/chemokine programs.
Hepatoma Cells Produce CXCL10 in Response to Type I IFN Stimulation.
We next asked whether hepatoma can produce IFN-inducible proteins in response to LPS indirectly via macrophages. The induction of IFN-β and IFN-γ was analyzed by RT-PCR (regular and quantitative, Fig. 5A and 5B, respectively). There was a significant increase in macrophage IFN-β expression, whereas IFN-γ remained undetectable (not shown). Next, we tested the effects of IFN-β on hepatoma cells. As shown in Fig. 4d, addition of IFN-β up-regulated CXCL10 expression (96-fold increase, as compared with untreated controls), and to a much less degree CXCL9/CXCL11 (5-fold to 6-fold), indicating that hepatoma cells responded to type I IFN by producing CXCL10. Addition of IFN-β also activated macrophages, as evidenced by increased expression of IFN-inducible proteins; that is, CXCL10 showed a greater than 400-fold increase, CXCL11 increased 70-fold, and CXCL9 increased 7-fold (Fig. 4C). Unlike in hepatoma cultures, IFN-β did up-regulate macrophage expression of TNF-α (15-fold) and CCL2 (6-fold) (Fig. 4A, B).
IFN-β–Dependent Interaction Between Hepatoma and Macrophages.
Although hepatoma cells failed to directly respond to LPS, they did respond to IFN-β, an LPS-activated macrophage product. Additionally, LPS-activated macrophages produced IFN-β, TNF-α, and IL-1β, all stimulatory to hepatoma. Thus, the question as to whether hepatoma cells may respond to LPS via macrophage-derived type I IFN was of particular importance. As shown in Fig. 6B, hepatoma readily produced CXCL10 in response to LPS when co-cultured in transwells with macrophages, or when stimulated with LPS-activated macrophage supernatants. Addition of neutralizing antibodies against IFN-β prevented CXCL10 induction (Fig. 6A). These results indicate that hepatoma cells responded to LPS indirectly when co-cultured with macrophages, and IFN-β mediated CXCL10 expression in LPS-stimulated hepatomas. Thus, IFN-β mediates potential synergy between the 2 cell types in their response to TLR4 activation.
In the first part this study, we found that IFNAR, but not IFNGR, deficient mice failed to elicit inflammation response and were protected from liver IRI. These results document, for the first time, that the type I but not the type II IFN pathway plays a key role in the induction of pro-inflammatory response and the development of liver IRI. Because TLR4 activation triggers local inflammation, the failure to sustain CXCL10, IL-6, and TNF-α gene induction in vitro provides the cellular basis for our in vivo findings in IFNAR KO mice. Having documented the key role of type I IFN signaling in the pathophysiology of liver IRI, we then asked whether our finding can be generalized and essentially broadened to well-defined liver cell-type substitutes, that is, hepatoma and macrophage lines and LPS-induced TLR4 responses in cell culture systems. We found that, unlike macrophages, which responded directly to TLR4 stimulation by producing pro-inflammatory cytokine/chemokines, hepatoma responded to LPS indirectly via macrophage-derived type I IFN and by up-regulating CXCL10. Our data also show that macrophages produced CXCL10 in response to LPS in both type I IFN pathway–independent and IFN pathway–dependent fashion, and the latter was required to sustain CXCL10; type II IFN was not involved. Thus, type I IFN pathway mediates a potential synergy between macrophages and hepatocytes in TLR4-triggered inflammation response.
Because type I IFN is the major pathway activated by the MyD88-independent signaling downstream of TLR4, the current data provide another important piece of evidence for our hypothesis that TLR4 and its IRF3 signaling pathway are critical for liver inflammation and hepatocellular damage.7 In addition, this study supports the role of type I IFN in a noninfectious disease model, which has been mostly associated with antiviral immune responses.16 Our findings also may help to clarify the controversial role of IFN-γ in IRI14, 17 by documenting that type I IFN was the only IFN induced early during reperfusion. This does not preclude, however, the role of type II IFN or other signaling pathways in the later IRI phase. Our data in IFNGR KO mice are consistent with reports in which IFN- deficient mice developed IR-mediated kidney and liver damage, comparable with WT.12 However, others have shown that pharmacological doses of IFN-γ may produce anti-inflammatory effects,18 consistent with our finding that INFRG KO mice suffered even more severe IRI damage than WT. Indeed, IFN-γ may exert cytoprotective functions by modulating the production of proteases/reactive oxygen species from KC,19 by decreasing CD5420 or E/P selectin21 expression. Although both IFNs stimulate the production of IFN-inducible proteins,22 our results have revealed that type I IFN selectively mediated liver CXCL10 induction.
Liver resident macrophages KCs are believed to play a key role in initiating the pro-inflammatory response against IR.1, 4, 23 Neutrophils, however, are generally considered the effector cells causing the direct hepatocellular damage.24, 25 Recently, T cells, particularly of the CD4 phenotype, have been revealed as the key players regulating KC and neutrophil function.13, 26 However, the mechanism of T cell function in IRI remains unclear. We have shown that the induction of T cell–targeted chemokine CXCL10 correlated with the development of liver inflammation in a murine nontransplant IRI model.7 In a rat orthotopic liver transplantation model, we have also observed a significant increase in hepatocyte CXCL10 expression. Moreover, targeting CXCR3, the CXCL10 receptor, ameliorated liver IRI and reduced T cell sequestration at the graft site.10 Thus, understanding the mechanism of CXCL10 induction is important to further elucidate the complex immunological mechanisms of T cell function in liver IRI. Although TLR4 activation on KC results directly in CXCL10 induction, our current data indicate that type I IFN is required for the full induction of CXCL10 in livers. Our in vitro study demonstrated that type I IFN was not only necessary for macrophages to sustain TLR4-triggered CXCL10 expression, but was also capable of stimulating CXCL10 production by hepatocytes.
Although TLR4 activation on KC (or macrophages) results directly in CXCL10 induction, it remains unclear whether the surrounding liver PC (hepatocytes) are also operational and actively contribute to liver inflammation. Because TLR4 transcripts have been found in hepatocytes,27 the obvious question arises whether hepatocytes may respond directly to TLR4 activation or indirectly to KC activation products. Our study shows that hepatoma cells responded indirectly to LPS-induced TLR4 activation via macrophage-derived IFN-β. The hepatocyte production of CXCL10 has been studied with stimulation ranging from stress of cell isolation process, exogenous IFN-γ to the mixture of TNF-α and IL-1β. Although type I IFN has been implicated in modulating CXCL10 because of the presence of conserved IFN-stimulated response elements in its gene promoter regions, the phenomenon has not been studied in physiological settings. Our study demonstrates that IFN-β is not only capable of, but also represents the key mediator in the macrophage activation products to stimulate CXCL10 production by hepatocytes. Considering the hepatocyte volume, it is reasonable to hypothesize that liver-derived CXCL10 contributes significantly to IRI. Indeed, our in vivo finding was supportive of this hypothesis, because livers in IFNAR KO mice were not only impaired in CXCL10 induction but also protected from IRI. Therefore, liver parenchyma cells may actively participate in IR-induced intrahepatic inflammation.
Several key issues warrant further investigation. First, hepatocytes, although expressing minimal TLR4 levels, may still be capable of directly responding to LPS by inducing distinctive set of cytokine genes under certain conditions. Because IFNAR KO mice reveal multiple immune deficiencies other than type I IFN response,28 our in vivo data did not provide direct evidence that the defect in the KC–hepatocyte interaction itself is only instrumental for liver cytoprotection in IFNAR KO mice. The WT bone marrow–reconstituted IFNAR KO chimeras should produce a more definitive proof. In addition, although leukocytes other than KC may engage in this interaction, especially in the later IRI phase, we believe our work provides compelling evidence that type I but not type II IFN signaling is required for liver IR early inflammation response. Second, we are aware that KC represent the unique macrophage population able to produce CXCL10 in response to LPS.29 Although our inability to extend the conclusion on LPS response from bone marrow macrophages directly to liver KC represents a limitation, it has recently been shown that KC do express functional TLR4 (including CD14 and MD2) and respond to LPS in a fashion similar to other macrophage populations.30, 31 Conversely, the use of in vitro macrophage/hepatoma cell culture systems broaden the significance of our findings and allowed study of TLR4-dependent events in well-defined liver cell substitutes rather than in native liver cells that have been exposed to prolonged and often harmful ex vivo separation procedures. Although macrophage–hepatoma cell culture experiments support our hypothesis, KC–hepatocyte interactions in vivo still need to be evaluated in future studies.
In conclusion, this study provides further insight on the mechanism of TLR4 signaling and reveals a novel role for type I IFN in a noninfectious liver IRI model. Our in vitro findings point to a potential synergy in the development of inflammation cascade in vivo, initiated by macrophage–TLR4 activation, and leading to active involvement of hepatocytes via type I IFN signaling. Although CXCL10 induction represents the major consequence of type I IFN activation, its impact on the induction of other proinflammatory gene programs in different liver cell types, as well as on liver cell growth and death, warrant further investigation. The distinct roles of type I versus type II IFN also may provide the rationale for novel and effective means to ameliorate liver IRI in the clinics.