Ischemia/reperfusion injury (IRI) represents the major problem in clinical liver transplantation. We have shown that toll-like receptor 4 (TLR4) signaling is specifically required in initiating antigen-independent IRI leading to liver inflammation, whereas local induction of anti-oxidant heme oxygenase-1 (HO-1) is cytoprotective. This study analyzes in vivo interactions between HO-1 and sentinel TLR system in the pathophysiology of liver IRI. Using a 90-min lobar warm ischemia model, wild type (WT), TLR4 KO/mutant and TLR2 KO mice were first assessed for the severity of hepatocellular damage at 6 h postreperfusion. Unlike in WT or TLR2-deficient mice, disruption/absence of TLR4 pathway reduced IRI, as manifested by liver function (serum alanine aminotransferase levels), histology (Suzuki's scores), neutrophil infiltration (myeloperoxidase activity) and local/systemic TNF-α production (mRNA/protein levels). Moreover, defective TLR4 but not TLR2 signaling increased mRNA/protein HO-1 expression. In contrast, tin protoporphyrin-mediated HO-1 inhibition restored hepatic damage in otherwise IRI-resistant TLR4 mutant/KO mice. CoPP-induced HO-1 overexpression ameliorated hepatic damage in IRI-susceptible TLR2 KO mice, comparable with WT controls, and concomitantly diminished TLR4 levels. In conclusion, this study highlights the importance of cross talk between HO-1 and TLR system in the mechanism of hepatic IRI. Hepatic IRI represents a case for innate immunity in which HO-1 modulates proinflammatory responses that are triggered via TLR4 signaling, a putative HO-1 repressor.
Ischemia/reperfusion injury (IRI), an Ag-independent inflammatory response, remains an important clinical consideration in liver surgical intervention, trauma and transplantation. Indeed, severe damage related to organ retrieval, preservation and reperfusion leads to primary nonfunction of the graft, and may adversely affect the development of chronic rejection (1,2). The mechanism of liver IRI involves a complex interaction of events, which include neutrophil release and sinusoidal endothelial cell death followed by hepatocyte damage (3). These destructive effects arise from the acute generation of reactive oxygen species subsequent to reoxygenation, which inflict direct tissue injury. Furthermore, the activation of leukocytes, especially CD4 T and Kupffer cells, can be seen shortly after reperfusion. Such activated Kupffer cells produce TNF-α, which in turn can lead to endothelial cell activation and death.
Heme oxygenase-1 (HO-1; heat shock protein-hsp32), a rate-limiting enzyme in the catabolism of heme, is among the most critical protective mechanisms activated during times of cellular stress, and it is thought to play a key role in maintaining anti-oxidant/oxidant homeostasis (4). Others and we have documented that HO-1 activation exerts potent adaptive anti-inflammatory functions in a number of pathophysiological conditions, including allograft rejection and Ag-independent IRI in transplant recipients (reviewed in 5–7). Our demonstration that HO-1 overexpression is a prerequisite for cytoprotection against hepatic IRI (8–10) supports the idea that HO-1 might function as a “therapeutic funnel” required for the action of several therapeutic molecules (5). Although we have identified macrophages and Kupffer cells as prime hepatic HO-1 producers (11), the molecular mechanisms that initiate host proinflammatory responses in hepatic IRI remain to be elucidated.
The evolutionarily conserved sentinel toll-like receptors (TLR) belong to the IL-1R family, and recognize bacterial/viral-specific pathogen-associated molecular patterns (12,13). TLRs trigger host inflammatory responses that are mediated by macrophages, neutrophils and complement. The induced cytokine mediators may then activate systemic responses to recruit leukocytes to sites of inflammation. We (3) and others (14) proposed that the primary non-specific injury (such as IRI) to the donor organ induces events of the two distinct host immune defense systems: (i) a more broadly directed innate defense (evolutionary directed against infections), resulting in (ii) a definite adaptive defense that leads ultimately to a specific graft injury (i.e. rejection). Thus, IRI creates a milieu of inflammation that operates as a 'danger' signal in transplant recipients, and in analogy to the infection-induced inflammation, it may initiate a state of innate immunity by activating host TLR system. Indeed, we have recently documented the selective functional usage of TLRs in a non-infectious disease model, by showing the requirement for TLR4 but not TLR2 to trigger liver IRI cascade (15). This study was designed to analyze putative cross talk interactions between HO-1 and TLR systems in the pathophysiology of hepatic IRI.
Materials and Methods
Male wild type (WT) (C3H/HeOuj, C57BL/10SnJ and C57BL/6), TLR4 mutant (C3H/HeJ; Pro712-His712) and TLR4-deficient (KO; C57BL/10ScCr) mice (8–12 weeks old) were used (Jackson Laboratory, Bar Harbor, ME, USA). TLR2 KO mice (C57BL/6) were provided by Dr. G. Cheng (UCLA). Mice were housed in the UCLA animal facility under specific pathogen-free conditions. All animals 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 by NIH (publication 86-23 revised 1985).
Liver IRI model
We have used a warm hepatic IRI model in mice, as described (8,9). Briefly, mice are injected with heparin (100 μg/kg), and an atraumatic clip is used to interrupt the arterial/portal blood supply to the cephalad-liver lobes. After 90 min of warm ischemia, the clip is removed initiating hepatic reperfusion. Mice are sacrificed at 6 h of reperfusion, and liver tissue/peripheral blood is collected. The extent/severity of hepatic IRI in this study was assessed in groups of WT, TLR4 mutant and TLR4/TLR2 KO mice. Sham C57BL/10SnJ WT controls underwent same procedures, but without vascular occlusion. Each experiment was repeated at least twice and the same tissue samples were analyzed throughout.
To elucidate the functional significance of HO-1, groups of TLR4 mutant and TLR4 KO mice were treated side-by-side with tin protoporphyrin (SnPP; Porphyrin Products, Logan, UT, USA), a competitive HO-1 inhibitor (16). SnPP was diluted in 100 mM NaOH to a stock solution of 50 mM and kept at −70°C. SnPP was administered twice (30 μM/kg i.p.), 1 day prior to the experiment (day 1), and at the time of hepatic ischemia (day 0). This protocol inhibited HO-1 in our previous studies (8,9). In addition, groups of TLR2 KO and WT mice were treated with cobalt protoporphyrin (CoPP; Porphyrin Products), an HO-1 inducer (11,17). CoPP, dissolved in 0.2 mol/L sodium hydroxide (pH 7.4) was administered (5.0 mg/kg i.p.) 24 h before the onset of ischemia.
Liver and hepatocellular damage
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. The histological severity of IRI was graded using Suzuki's criteria (18), with modifications. In this classification, sinusoidal congestion, hepatocyte necrosis and ballooning degeneration are graded from 0 to 4. No necrosis, congestion/centrilobular ballooning is given a score of 0, whereas severe congestion/ballooning degeneration, as well as >60% lobular necrosis are given a value of 4.
Serum alanine aminotransferase (sALT) levels, an indicator of hepatocellular injury, were measured in blood samples obtained 6 h after hepatic reperfusion. Measurements of sALT were made using an auto analyzer by ANTECH Diagnostics (Los Angeles, CA, USA).
The presence of myeloperoxidase (MPO) was used as an index of neutrophil accumulation (19). Briefly, liver tissue was placed in 0.5% hexadecyltrimethyl-ammonium bromide and 50-mMol potassium phosphate buffer solution (pH = 5.0). Each sample was homogenized, centrifuged, and supernatants were mixed with hydrogen peroxide-sodium acetate and tetramethyl-benzidine solutions. The change in absorbance was measured by spectrophotometer at 655 nm. One unit (AU) of MPO activity was defined as the quantity of enzyme degrading 1 μMol peroxide/min/g at 25°C.
To study TNF-α and HO-1 gene expression, we used competitive-template RT-PCR, as described (20). Briefly, after extracting total RNA from liver tissue, 5 μg of RNA was reverse-transcribed into cDNA using oligo (dT) primers with superscript reverse transcriptase (Life Technologies, NY, USA). PCR products were electrophoresed in ethidium bromide-stained 2% agarose gel, and scanned by using Kodak Digital Science 1D Analysis Software (Version 2.0). To compare the relative level of each cytokine, all samples were normalized against the respective β-actin template cDNA.
We used ELISA kit (eBioscience, San Diego, CA, USA) to measure serum TNF-α levels (10). Briefly, capture Ab was coated onto 96 well EIA plate (Costar, Corning, NY, USA), serum samples diluted at 1:5 were duplicated and the standard was used at 1000 pg/mL, followed by 1–2 serial dilutions. Biotinylated detecting Abs were added followed by avidin-HRP. TMB was used as the substrate and the color reaction was stopped by 1M H3PO4. Serum concentrations of TNF-α were calculated from the standard curve.
Extracted protein from liver samples (30 μg/sample) was subjected to 12% SDS-polyacrylamide gel electrophoresis (PAGE). The gel was stained with coomassie blue to document equal protein loading. Membrane was blocked with 3% dry milk/0.1% Tween 20 (USB, Cleveland, OH, USA) and incubated with rabbit anti-mouse HO-1 Ab (Stressgen Biotechnology, Canada) or rabbit anti-mouse TLR2/TLR4 Ab (Santa Cruz Biotechnology, Santa Cruz, CA, USA). The filters were incubated with horseradish peroxidase donkey anti-rabbit Ab (Amersham, Arlington Heights, IL, USA). Relative quantities of HO-1, TLR2 and TLR4 proteins were determined using a densitometer (Kodak Analysis Software, Rochester, NY, USA).
All values are expressed as mean ± SD. Data were first analyzed using one-way ANOVA, and individual groups were subsequently compared with unpaired two-tailed Student's t-test. P values of less than 0.05 were considered statistically significant.
Disruption of TLR4 signaling reduces hepatic IRI
Ninety minutes of hepatic warm ischemia followed by 6 h of reperfusion significantly increased sALT levels (IU/L) in WT mice, as compared with sham-operated controls (2348 ± 809 and 2149 ± 485 vs. 27 ± 4, respectively; p < 0.01; Figure 1). In agreement with our recent findings (15), sALT levels were reduced in TLR4 mutant and TLR4 KO mice (285 ± 75 and 269 ± 37, respectively; p < 0.05). In contrast, sALT levels in TLR2 KO mice exposed to hepatic IRI remained elevated, comparable with WT controls (2255 ± 717 and 2992 ± 834, respectively).
These data correlated with histological Suzuki's criteria of hepatic IRI. As shown in Figure 2A,F, livers in WT and TLR2 KO mice revealed severe lobular distortion, edema, centrilobular ballooning and sinusoidal congestion with hepatocellular necrosis of 20–50% (Suzuki's score = 4.5 ± 0.6 and 4.0 ± 0.7, respectively). In contrast, livers in TLR4 mutant (Figure 2B) and TLR4 KO (Figure 2D) mice showed good preservation of lobular architecture without edema, ballooning or necrosis (score = 1.1 ± 0.3 and 1.0 ± 0.4, respectively).
Local neutrophil accumulation in ischemic lobes, as analyzed by MPO activity (IU/g), increased from 0.4 ± 0.1 in sham-operated controls to 4.7 ± 0.3 and 4.6 ± 0.4 in WT groups (Figure 3; p < 0.01). The disruption of TLR4 signaling in TLR4 mutants or its absence in TLR4 KOs reduced MPO activity to 2.5 ± 0.3 and 1.9 ± 0.3, respectively (p < 0.05, as compared with WT). In contrast, MPO activity in TLR2 KO mice undergoing IRI remained elevated, comparable with WT group (4.6 ± 0.4 and 4.8 ± 0.3, respectively).
Defective TLR4 signaling diminishes TNF-α production
We used competitive-template RT-PCR to analyze the expression of mRNA coding for TNF-α, a key cytokine in the pathophysiology of hepatic IRI (1). As shown in Figure 4A, livers in WT mice exposed to warm ischemia showed increased expression of mRNA coding for TNF-α, as compared with sham controls (2.3 ± 0.1, 2.3 ± 0.1 and 2.3 ± 0.1 vs. 0.2 ± 0.03, p < 0.005). TNF-α expression in TLR4 mutant and TLR4 KO mice was diminished, as compared with WT groups (1.9 ± 0.2 and 1.5 ± 0.1, respectively; p < 0.05). In contrast, TNF-α levels in TLR2 KO mice remained elevated, comparable with WT controls (2.3 ± 0.2 and 2.3 ± 0.1, respectively).
Since most of TNF regulation is posttranscriptional, we next assessed TNF-α serum protein levels. As shown in Figure 4B, ELISA-assisted expression of TNF-α (ng/mL) in TLR4 mutant or TLR4 KO mice (99.7 ± 9.7 and 42.8 ± 8.6, respectively) was significantly lower (p < 0.01), as compared with WT (641 ± 67 and 626 ± 60, respectively). Unlike disruption of TLR4 signaling, TNF-α levels remained elevated in TLR2 KOs, comparable with corresponding WT controls (617 ± 53 and 631 ± 77, respectively).
Defective TLR4 but not TLR2 signaling increases HO-1 expression
We used competitive-template RT-PCR to analyze the expression of HO-1, an anti-oxidative hsp32 with cytoprotective functions (4–7). Indeed, as shown in Figure 5 livers in TLR4 mutant or TLR4 KO mice showed increased expression of mRNA coding for HO-1 (1.6 ± 0.2 and 1.8 ± 0.2), as compared with respective WT controls (0.6 ± 0.2; p < 0.05, and 0.5 ± 0.1; p < 0.02) or TLR2 KO recipients (0.5 ± 0.1).
To confirm the translation of mRNA into the protein, we evaluated the expression of HO-1 as well as TLR4/TLR2 in hepatic ischemic lobes by Western blots. The relative expression levels in absorbance units (AU) were analyzed by densitometry. As shown in Figure 6, improved liver function in TLR4 mutant and TLR4 KO mice resulted in consistently enhanced expression of HO-1 (line 3 and 4, respectively; 1.8 ± 0.1 AU). In contrast, the corresponding samples in WT mice showed reduced HO-1 levels (line 2; 0.3 ± 0.3 AU). HO-1 expression in TLR4 competent but TLR2-deficient recipients remained diminished, comparable with WT (line 8; 0.2 ± 0.1 AU) or sham (line 1; 0.1 ± 0.07 AU) controls.
HO-1 inhibition restores hepatic IRI in TLR4 mutant/KO mice
To analyze as to whether or not HO-1 overexpression might exert cytoprotective function despite defective TLR4 signaling, groups of TLR4 mutant and deficient mice were pretreated side-by-side with SnPP, an HO-1 inhibitor (16). Depression of HO-1 mRNA (Figure 5; 0.5 ± 0.15 and 0.7 ± 0.1, respectively) and HO-1 protein (Figure 6, line 6; 0.5 ± 0.1 AU) restored hepatic damage in otherwise IRI-resistant TLR4 mutant/KO mice, as evidenced by increased sALT levels (IU/L; 2301 ± 746 and 622 ± 89 vs. 285 ± 75 and 269 ± 37; p < 0.05 and p < 0.01, respectively, Figure 1) and enhanced MPO activity (U/gm; 4.5 ± 0.4 and 3.4 ± 0.3 vs. 2.5 ± 0.3 and 1.9 ± 0.3, respectively; p < 0.05; Figure 3). Moreover, SnPP-induced HO-1 depression augmented local TNF-α mRNA (2.24 ± 0.18 and 1.93 ± 0.11, respectively; p < 0.05; Figure 4A) and systemic TNF-α protein (485.6 ± 75.0 and 433.0 ± 21.8, respectively; p < 0.02; Figure 4B) levels. Histological damage in SnPP-treated TLR4 mutant mice (Figure 2C) was comparable with WT counterparts, as evidenced by hepatocellular necrosis of ca. 20–50% (Suzuki's score = 4.0 ± 0.5). Somewhat less necrosis (5–10%) was found in SnPP-treated TLR4 KO livers (Figure 2E; score = 3.4 ± 0.4). SnPP treatment was not hepatocytotoxic on its own, as sALT levels and liver histology in naïve mice treated with SnPP remained normal (data not shown).
HO-1 induction prevents hepatic injury in TLR2 KO mice
Finally, we asked as to whether or not HO-1 overexpression might protect TLR2 KO mice from fulminant hepatic IRI. Groups of TLR2 KO and WT mice were pretreated with CoPP, an HO-1 inducer (11,17). Indeed, the induction of HO-1 mRNA (Figure 5; 1.94 ± 0.21) and protein (Figure 6, line 9; 2.3 ± 0.1 AU) ameliorated hepatic damage in otherwise IRI-susceptible TLR2 KO mice. This was evidenced by depressed sALT levels (IU/L; 542 ± 205 vs. 2255 ± 717; p < 0.005, Figure 1), and decreased MPO activity (U/gm; 1.88 ± 0.15 vs. 4.6 ± 0.4; p < 0.05, Figure 3), as compared with untreated TLR2 KO recipients. The hepatic histology in CoPP-treated TLR2 KO mice showed preservation of lobular architecture without edema, ballooning or necrosis (n = 3/group; data not shown). In addition, CoPP-triggered HO-1 overexpression in TLR2 KO mice was accompanied by diminished TLR4 protein levels (Figure 6, line 9; 0.4 ± 0.1 AU). The effects of CoPP treatment in WT mice were comparable with those in TLR2 KO recipients (sALT levels = 542 ± 174 IU/L, n = 3; data not shown).
This study demonstrates that warm hepatic ischemia/reperfusion resulted in equally severe hepatocellular injury in TLR2 KO and WT mice, accompanied by enhanced synthesis of TNF-α and expression of TLR4 protein. In contrast, disruption of TLR4 signaling or its absence in TLR4 mutant and TLR4 KO mice, respectively, ameliorated hepatic insult and reduced TNF-α levels without affecting TLR2 expression. These beneficial effects following TLR4 blockade were dependent on intrahepatic overexpression of HO-1 mRNA/protein, data supported by studies in which HO-1 induction prevented liver damage and downregulated TLR4 in IRI-susceptible TLR2 KO mice. This study highlights the importance of cross talk between HO-1 and TLR system in the mechanism of liver IRI.
Although eleven different mammalian TLRs have been identified by sequence analysis, only TLR4 is considered the major receptor for lipopolysaccharide (LPS)-induced infectious responses (21). However, TLR4 activation has been implicated in the mechanism of non-infectious intimal lesion causing arterial obstructive disease (22) and murine myocardial IRI (23), whereas TLR2 has been involved in modulating ventricular remodeling after myocardial infarction (24). In agreement with our own (15) and others (25) findings, we first confirmed the involvement of TLR4 signaling in hepatic IRI. Indeed, 90 min of ischemia followed by 6 h of reperfusion significantly increased TLR4 as well as TNF-α expression, neutrophil accumulation and induced severe liver injury in WT mice. However, disruption of TLR4 signaling (in TLR4 mutant/TLR4 KO mice) ameliorated hepatic IRI. Previous in vitro studies have shown that macrophages from TLR4 mutant and KO mice failed to induce TNF-α, IL-1 and IL-6 (26). However, adenoviral murine TLR4-transduced hepatocytes displayed increased NF-κB activation (27). Collectively, TLR4 blockade affects the function of hepatocytes and Kupffer cells, the resident macrophages of the liver, depresses the production of proinflammatory cytokines and ameliorates hepatocellular IRI.
TLR4 can recognize several endogenous ligands, the best of which characterized are heparan sulfate (28), hyaluronan (29), fibronectin extra domain A (30), fibrinogen (31) and heat shock proteins 60/70 (32–34). These TLR4 ligands are released from necrotic cells and may induce DC maturation by activating NF-κB (32), and thus initiating type 1-dependent tissue inflammation (35). Their degraded fragments during tissue injury have been shown to exert proinflammatory activity through TLR4. The prominent hepatocellular necrosis in the ischemic lobes of WT mice might trigger HSPs from necrotic cells and provoke an inflammatory response. Ohashi et al. (33) reported that stimulation of TNF-α response by HSP60 was dependent on the presence of a functional TLR4, whereas macrophages from mutant TLR4 failed to respond to HSP60. These findings are in agreement with our present data. Indeed, the production of TNF-α, a TLR4 activation marker, was reduced in TLR4 mutant/KO mice, suggesting that putative TLR4 endogenous ligands do play a role in the mechanism of liver IRI. We are aware that increased LPS levels have been recorded in liver transplant recipients (36), possibly resulting from increased endotoxin translocation from the gut due to a decrease of intestinal/hepatic blood flow. However, it is doubtful that gut-derived LPS initiates IRI, as liver damage due to ischemia can develop ex vivo following blood perfusion and independent of intestinal/hepatic blood flow (11,17). Thus, increased LPS levels in circulation resulting from intestinal damage are most likely secondary to the initial IRI.
There were no significant differences in hepatocellular injury, neutrophil accumulation and TNF-α/TLR4 expression between TLR2 KO and WT controls. This may imply that TLR2 signaling does not play an important role in hepatic IRI. Heine et al. (37) showed that Chinese hamster ovary fibroblasts with a truncated non-functional TLR2 respond to LPS. Takeuchi et al. (38) demonstrated that TLR2 KO mice die on lethal challenge with LPS and their macrophages produce cytokines comparable with WT littermates. These reports are consistent with our findings on the lack of cytoprotection in TLR2 KO mice. However, a possible role of TLR2 as a helper receptor for TLR4 cannot be ruled out. The regulation of TLR2 and other members of the TLR family may also influence early proinflammatory responses in hepatic IRI.
Our findings document the importance of in vivo cross talk interactions between TLR2, TLR4 and HO-1 signaling networks. Indeed, diminished HO-1 in both mRNA/protein levels was consistently observed in ischemic hepatic lobes of WT and TLR4 competent TLR2 KO mice. In contrast, 'cytoprotected' ischemic lobes of TLR4 mutant/KO mice all strongly overexpressed HO-1. Our previous studies (11) have identified Kupffer cells as the prime HO-1 producers in the mouse liver. Although in our ongoing RT-PCR studies, we readily detect mRNA coding for TLR4 in mouse liver subjected to IRI, macrophages but not hepatocytes respond to LPS stimulation in the culture (Zhai, unpublished data). This implies that TLR4 activation on Kupffer cells rather than hepatocytes is responsible for innate immune activation. Unlike in vitro (4,5), but consistent with our previous in vivo studies (8,9), SnPP treatment consistently decreases HO-1 mRNA/protein expression in mouse livers. Consequently, SnPP-facilitated HO-1 depression recreated hepatic IRI in both TLR4 mutant and TLR4 KO mice, as assessed by liver histology and MPO activity. However, the hepatocellular damage, as assessed by sALT levels after SnPP treatment, was more pronounced in TLR4 mutant than in TLR4 KO mice. This might be explained by different genetic animal backgrounds and/or putative in vivo hyporesponsiveness after TLR4 genetic disruption to endogenous ligands, as opposed to TLR4 deletion. It is plausible that IRI (measured by sALT) might requite low TLR4 activity that is present in TLR4 mutant but not TLR4 KO mice. This indicates that HO-1 overexpression counters IRI-induced hepatocellular damage. As SnPP-mediated HO-1 depression exacerbated MPO activity in both TLR4 KO and mutant mice, neutrophil infiltration/activation, unlike hepatocellular damage (assessed by sALT), seems to be TLR4 independent. Indeed, despite similar neutrophil sequestration, higher liver injury occurred only when TLR4 signaling was not ablated (i.e. in TLR4 mutants). Collectively, our results document that HO-1 has an essential function in TLR4-mediated signaling. In parallel, CoPP-mediated HO-1 mRNA/protein induction ameliorated hepatic damage and diminished TLR4 signaling in otherwise IRI-susceptible TLR2 KO as well as WT mice. All these support the key role of HO-1 in preventing host inflammatory responses and suggest that TLR4 might function as putative HO-1 repressor in non-infectious hepatic IRI. Others have shown that TLR4 plays a dominant role in mediating HO-1 expression in LPS-infused mouse liver (39). Hence, hepatic IRI represents a case for innate immunity in which HO-1 modulates proinflammatory responses triggered via TLR4. The complex molecular mechanisms by which the cross talk between sentinel TLR4–TLR2 system and anti-oxidant HO-1 may influence Ag-independent proinflammatory IRI sequel remain to be elucidated.
This study was supported by NIH Grants RO1 DK062357, AI23847, AI42223 (JWKW), The Roche Organ Transplantation Research Foundation (YZ) and The Dumont Research Foundation.