Heme oxygenase (HO-1) provides a cellular defense mechanism during oxidative stress and catalyzes the rate-limiting step in heme metabolism that produces biliverdin (BV). The role of BV and its potential use in preventing ischemia/reperfusion injury (IRI) had never been studied. This study was designed to explore putative cytoprotective functions of BV during hepatic IRI in rat liver models of ex vivo perfusion and orthotopic liver transplantation (OLT) after prolonged periods of cold ischemia. In an ex vivo hepatic IRI model, adjunctive BV improved portal venous blood flow, increased bile production, and decreased hepatocellular damage. These findings were correlated with amelioration of histological features of IRI, as assessed by Suzuki's criteria. Following cold ischemia and syngeneic OLT, BV therapy extended animal survival from 50% in untreated controls to 90% to 100%. This effect correlated with improved liver function and preserved hepatic architecture. Additionally, BV adjuvant after OLT decreased endothelial expression of cellular adhesion molecules (P-selectin and intracellular adhesion molecule 1), and decreased the extent of infiltration by neutrophils and inflammatory macrophages. BV also inhibited expression of inducible nitric oxide synthase and proinflammatory cytokines (interleukin 1β, tumor necrosis factor α, and interleukin 6) in OLTs. Finally, BV therapy promoted an increased expression of antiapoptotic molecules independently of HO-1 expression, consistent with BV being an important mediator through which HO-1 prevents cell death. In conclusion, this study documents and dissects potent cytoprotective effects of BV in well-established rat models of hepatic IRI. Our results provide the rationale for a novel therapeutic approach using BV to maximize the function and thus the availability of donor organs. (HEPATOLOGY 2004;40:1333–1341.)
Ischemia and reperfusion injury (IRI), an antigen-independent component of the insult to the liver during “harvesting,” represents an important problem affecting liver transplantation. IRI causes up to 10% of early liver failures and can lead to a higher incidence of acute and chronic rejection.1 Moreover, with the increasing donor shortage, more functionally “suboptimal” or “marginal” livers are being used. Such livers are more susceptible to the damage caused by IRI compared with normal livers.2 Indeed, minimizing the adverse effects of IRI could significantly increase the number of patients that successfully undergo liver transplantation.
Liver IRI is mediated by several processes that lead to hepatocellular damage, which is triggered when the liver is transiently deprived of oxygen during the organ procurement for transplantation, and later reoxygenated during reperfusion. The structural changes promoted by cold ischemia and reperfusion become more prominent with increased storage time.3 The sinusoidal endothelial cells are very sensitive to IRI, thus affecting the delicate balance that maintains homeostasis in the microcirculation with attraction, activation, adhesion, and migration of polymorphonuclear neutrophils (PMNs). These events cause local tissue destruction through the release of proteases and oxygen-free radicals. Oxygen-free radical–induced injury targets proteins, enzymes, nucleic acids, the cytoskeleton, cell membranes, and lipid peroxides, resulting in decreased mitochondrial function and lipid peroxidation.4
Heme oxygenases (HOs) are ubiquitous enzymes that catalyze the initial and rate-limiting steps in the oxidative degradation of heme into biliverdin (BV), carbon monoxide, and free iron. BV is reduced to bilirubin (BR) by bilirubin reductase, and the free iron is used in intracellular metabolism or sequestered into ferritin. It is believed that the products derived from the catalysis of heme by HO—namely BV (in addition to the generated BR) and carbon monoxide—can provide the physiological effects of HO. In addition, the upregulation of ferritin by the free iron has similar effects.5 The upregulation of HO-1 is thought to be a protective response from cellular stress following ischemia, inflammation, and radiation, preventing the deleterious effects of heme, which promotes lipid peroxidation and free radical formation as well as mediating the anti-inflammatory and antiapoptotic effects via its products. Because activation of HO-1 affords a cytoprotective function, its use as a novel strategy to prevent IRI has been extensively studied.6 In the clinical setting, however, upregulating HO-1 may be a suboptimal approach for many reasons. First, no known reagents specifically induce HO activity. Second, individuals vary in the extent of their HO-1 response to a given stress due to a promoter polymorphism.7 Third, there may be undesirable effects of treatment with nonspecific HO-1 inducers.8 In this context, determining if BV—which can be readily administered at a given dose and dosing schedule—can be used to prevent hepatic IRI is potentially important.
This study was designed to determine the role of BV treatment in well-defined rat liver models of prolonged cold ischemia followed by ex vivo reperfusion or syngeneic orthotopic liver transplantation (OLT). Here, we demonstrate that a nontoxic BV regimen can ameliorate severe liver IRI independent of HO-1 expression. The beneficial effects of adjuvant BV treatment on hepatocyte injury and liver function were comparable with those seen in parallel with HO-1 upregulation, which points to BV as one of the mainstream HO-1 mediators in IRI protection. Our findings provide the rationale for a novel therapeutic approach using BV to maximize the organ donor pool through the safer use of liver transplants despite prolonged periods of cold ischemia.
Liver Cold Ischemia Model Followed by Ex Vivo Reperfusion.
Sprague-Dawley rats (250-300 g; Harlan Sprague-Dawley, Inc., Indianapolis, IN) underwent isoflurane anesthesia and systemic heparinization. After skeletonization of the liver, the portal vein, bile duct, and inferior vena cava were cannulated and the liver was flushed with 10 mL of University of Wisconsin (UW) solution. Livers were stored for 24 hours at 4°C in UW, then perfused with whole blood for 2 hours on an isolated rat liver perfusion apparatus with stable temperature (37°C), pressure (13 cm H2O), and pH (7.3) as described.9
Biliverdin dihydrochloride (ICN Biomedicals, Inc., Aurora, OH) was dissolved in 2 mmol/L NaOH and kept in the dark. BV was added to the perfusate (10 or 50 μmol/L) or was administered to prospective liver donors 24 hours prior to the liver harvest (50 μmol/kg intravenously).
Portal vein blood flow and pressure were recorded every 15 minutes, while bile output was monitored every 30 minutes. Blood was collected at 30-minute intervals for serum glutamic-oxoaloacetic transaminase (sGOT) and serum glutamic-pyruvic transaminase levels, which were measured with an autoanalyzer (ANTECH Diagnostics, Irvine, CA). After 2 hours of perfusion, liver tissue was fixed in formalin for histological evaluation.
Liver Cold Ischemia Model Followed by Syngeneic OLT.
Syngeneic OLTs were performed using Sprague-Dawley livers stored for 24 hours at 4°C in UW solution prior to being transplanted into syngeneic Sprague-Dawley recipients with revascularization without hepatic artery reconstruction.10 There were two major experimental groups. In the BV group, livers were harvested from untreated rats, stored for 24 hours at 4°C, then transplanted into syngeneic recipients that were treated with BV (50 μmol/kg intravenously) immediately before and 20 hours after reperfusion. In the BBV group, recipients received the same treatment as in the BV group, plus donor rats were conditioned with BV (50 μmol/kg intravenously) as well.
The presence of myeloperoxidase, an enzyme specific for neutrophils, was used as an index of PMN accumulation in the liver as in our previous studies.11 One unit of myeloperoxidase activity was defined as the quantity of enzyme degrading 1 μmol peroxide/min at 25°C/g of tissue.
Liver specimens were fixed in 10% buffered formalin solution and embedded in paraffin. Sections were made at 4 μm and stained with hemotoxylin-eosin. The histological severity of IRI was graded using modified Suzuki's criteria.12 In this classification, sinusoidal congestion, hepatocyte necrosis, and ballooning degeneration are graded on a scale of 0 to 4. No necrosis, congestion, or centrilobular ballooning is given a score of 0, while severe congestion/ballooning degeneration, and more than 60% lobular necrosis is given a value of 4.
Liver samples were embedded in freezing medium and stored at −80°C. Five-micrometer sections were fixed in cold acetone and processed for immunohistochemistry using the peroxidase staining method.13 Primary antibodies included P-selectin (CD62P), intracellular adhesion molecule 1 (ICAM-1) (1A29), and ED-1 (1C7) (all from Pharmingen, San Diego, CA), and HO-1 (StressGen Biotech, Victoria, BC, Canada).
Western Blot Analysis.
Protein was extracted from liver samples with PBSTDS (50 mmol/L Tris, 150 mmol/L NaCl, 0.1% SDS, 1% sodium deoxycholate, and 1% Triton X-100; pH 7.2) buffer. Proteins (40 μg/sample) in sodium dodecyl sulfate–loading buffer were subjected to 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred to nitrocellulose membrane (Bio-Rad, Hercules, CA). The gel was stained with Coomassie blue to document equal protein loading. The membrane was incubated with specific primary antibody against HO-1 (StressGen Biotech), Bcl-2, Bag-1, caspase 3, and actin (Santa Cruz Biotechnology, Santa Cruz, CA). The relative quantities of proteins were determined with a densitometer and expressed in comparison with actin expression (Kodak Digital Science 1D Analysis Software, Rochester, NY).
RNA was extracted from livers with Trizol (Life Technologies, Inc., Grand Island, NY) using a Polytron RT-3000 (Kinematica AG, Littau-Luzern, Switzerland) as described.14 Reverse transcription was performed using 4 μg of total RNA in a first-strand complementary DNA synthesis reaction with SuperScript II RNaseH Reverse Transcriptase (Life Technologies, Inc.). One microliter of the resulting reverse transcriptase product was used for polymerase chain reaction amplification.
In Vivo Detection of Apoptosis.
Apoptosis was detected using the terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) method. A commercial in situ histochemical assay (Klenow-FragEL, Oncogene Research Products, Cambridge, MA) was performed to detect the DNA fragmentation in formalin-fixed paraffin-embedded OLT sections.15 The results were scored semiquantitatively by averaging the number of apoptotic cells per microscopic field at a magnification of ×200. Six fields were evaluated per tissue sample.
Results are expressed as the mean ± SD. Statistical comparisons between the groups were performed via Student t test using a Systat version 11 statistical package (SPSS Inc., Chicago, IL). A P value of less than .05 was considered statistically significant.
Liver Cold Ischemia Followed by Ex Vivo Reperfusion.
To test our hypothesis that BV decreases IRI-mediated hepatocyte damage, we monitored portal vein blood flow and resistance, bile production, and transaminase levels in Sprague-Dawley rat livers that were subjected to 24 hours of cold storage and then perfused for 2 hours with or without BV (10 or 50 μmol/L). Addition of BV at either concentration significantly improved portal blood flow compared with untreated controls (Fig. 1A). As shown in Fig. 1B, administration of BV significantly increased bile production. IRI-induced hepatocellular damage as measured by sGOT release was also markedly reduced in BV-treated livers (Fig. 1C).
Having shown that BV at the time of reperfusion exerts cytoprotection against IRI, we next investigated whether donor treatment with BV 24 hours prior to the liver harvest might exert similar beneficial effects. However, unlike livers treated with BV during reperfusion, BV pretreatment alone did not improve portal blood flow (see Fig. 1A) or bile production (see Fig. 1B). In addition, hepatocyte injury was comparable to controls (see Fig. 1C). Some of the livers perfused with BV at a dose of 50 μmol/L were also pretreated with BV 24 hours prior to the harvest (50 μmol/kg intravenously). These livers did not show any differences in portal blood flow, bile production, or hepatocyte function compared with the BV-treated group during reperfusion alone (see Fig. 1A–C).
The hepatocyte injury was graded using Suzuki's criteria.12 In the untreated IRI group, there was severe sinusoidal/vascular congestion with marked vacuolization focally associated with hepatocyte necrosis (Fig. 2A; score = 6.67 ± 0.82). In marked contrast, livers perfused with BV showed significant preservation of the lobular architecture with minimal signs of hepatocyte necrosis (Fig. 2B; score = 3.73 ± 1.1; P < .0001).
Liver Cold Ischemia Followed by Syngeneic OLT.
We next examined whether administration of BV could confer protection against hepatic IRI in an in vivo setting. Hence, we performed OLTs using livers that were stored at 4°C for 24 hours before being transplanted into syngeneic rat recipients. The treatment group received BV (50 μmol/kg intravenously) immediately before reperfusion and 20 hours after reperfusion. As shown in Fig. 3A, untreated recipients of OLTs stored in the cold had a 50% survival rate at day 7 posttransplant (6 of 12). In contrast, OLT recipients treated with BV showed 90% survival (9 of 10; P < .05). Treatment of both donors and recipients with BV resulted in 100% survival (10 of 10; see Fig. 3A). Prolonged survival after BV treatment correlated with markedly improved liver function, as evidenced by sGOT, serum glutamic pyruvic transaminase, and lactate dehydrogenase levels (Fig. 3B).
As shown in Fig. 2C, central vein congestion and extensive areas of necrosis with PMN infiltration adjacent to necrotic tissue characterized untreated OLTs. In contrast, BV-treated OLTs (Fig. 2D) showed mild changes without significant vascular congestion or necrosis (Suzuki's score at day 1 posttransplant = 2.68 ± 0.42 vs. 1.24 ± 0.86, respectively; P < .0001).
Figure 4A-B shows immunohistochemical staining for P-selectin (CD62), a molecule that mediates initial adhesive interactions of PMNs and monocytes with endothelium in inflammatory states. At 24 hours, untreated OLTs were characterized by high expression of P-selectin by venular endothelial cells compared with the BV-treated group, in which there was little if any expression (+++ vs. 0/+). Similarly, ICAM-1 was strongly upregulated in untreated compared with BV-treated OLTs (+++ vs. +; Fig. 4C–D).
To determine whether BV treatment affected local leukocyte infiltration, we assessed PMN infiltration using a myeloperoxidase assay and mononuclear cell infiltration by immunohistology. Myeloperoxidase activity in BV-treated OLTs was reduced at 6 hours (1.63 ± 0.41 vs. 0.96 ± 0.12; P < .01) and at 24 hours (2.18 ± 0.34 vs. 1.21 ± 0.28; P < .03) after transplantation compared with untreated OLTs (data not shown). In addition, macrophage infiltration of BV-treated OLTs was profoundly decreased compared with respective untreated controls (+ vs. +++/++++; Fig. 4E–F).
As shown in Fig. 5, untreated OLTs were characterized by elevated levels of inducible NO synthase (iNOS) messenger RNA at day 1 posttransplant compared with BV-treated OLTs (3.7 ± 1.2 vs. 1.1 ± 0.1; P < .001). BV treatment significantly reduced the expression of tumor necrosis factor α (TNF-α) (3.1 ± 0.9 vs. 1.6 ± 0.2; P < .01), interleukin (IL)-1β (1.4 ± 0.1 vs. 0.8 ± 0.1; P < .01) and IL-6 (0.8 ± 0.2 vs. 0.3 ± 0.1; P < .002).
Untreated OLTs showed significant hepatocellular apoptosis, as characterized by dense nuclear margination at 6 hours and 24 hours (12.5 ± 4.5 and 24.5 ± 5.0 TUNEL+ cells/field, respectively; Fig. 6A–B). In contrast, BV treatment profoundly diminished the number of apoptotic cells in OLTs (2.5 ± 1 and 3.3 ± 1.2 TUNEL+ cells/field at 6 hours and 24 hours, respectively; P < .005 and P < .0002; Fig. 6C–D).
Western blot analysis for the relative expression levels of proapoptotic (caspase 3) and antiapoptotic (Bcl-2/Bag-1) gene products was determined by densitometry and expressed as ratios with the expression of actin as a housekeeping gene. As shown in Fig. 7, BV treatment strongly downregulated caspase 3 expression at 6 hours and 24 hours (0.55 ± 0.03 and 0.54 ± 0.05, respectively) compared with untreated OLTs (0.83 ± 0.05 and 0.9 ± 0.06, respectively; P < .0001). Expression of antiapoptotic Bcl-2 and Bag-1 was enhanced in BV-treated OLTs at 6 hours compared with untreated OLTs (0.79 ± 0.07 and 0.69 ± 0.04, respectively) and at 24 hours (0.86 ± 0.1 and 0.74 ± 0.05, respectively) compared with untreated OLTs (Bag-1: 0.46 ± 0.05 and 0.48 ± 0.04, P < .0004; Bcl-2: 0.37 ± 0.07 and 0.41 ± 0.09, P < .0002). HO-1 expression in the BV group was downregulated at 6 hours and 24 hours (0.51 ± 0.05 and 0.63 ± 0.04) when compared with untreated OLTs (0.9 ± 0.08 and 1.01 ± 0.07; P < .0001) (see Fig. 7). The immunohistochemical staining for HO-1 also revealed the downregulation of HO-1 expression in BV-treated OLTs (+++ vs. +; Fig. 4G and 4H).
Induction of HO-1 affords important cytoprotective functions in hepatic IRI6, 8, 9, 11, 15–17; however, the role of BV—one of the key products of HO-1—and its potential use in preventing IRI have not been explored. We have demonstrated that BV treatment exerts striking cytoprotection in rat liver models of hepatic cold ischemia followed by ex vivo reperfusion or syngeneic OLT. Adjuvant BV treatment after OLT (1) improved liver function and preserved hepatocyte integrity with resultant prolongation of animal survival; (2) decreased the expression of cellular adhesion molecules and prevented mononuclear cell infiltration; (3) inhibited iNOS and proinflammatory cytokine expression; and (4) upregulated antiapoptotic molecules while downregulating HO-1.
The improvement of the portal vein blood flow, the increase of bile production, and the reduction of sGOT levels reflected the beneficial effects of BV treatment in the ex vivo IRI model. Resistance in the graft caused by lobular ballooning, hepatocyte swelling, and sinusoidal congestion mostly affects portal blood flow. The improved portal venous blood flow represents the lesser amount of hepatocyte injury and lobular disarray in the liver rather than the endothelium-dependent carbon monoxide vasodilatory effect.16 In the in vivo OLT model, we chose 24 hours of cold ischemia to accentuate the acute oxidative stress on reperfusion, as seen in the decreased bile production, lower portal vein blood flow, and higher hepatocellular injury compared with unpreserved livers. This model was accompanied by a 50% survival rate in the control group. BV treatment improved animal survival to 90% to 100% (P < .05), the ultimate test for liver function. Collectively, these results are consistent with the ability of BV to protect livers from IRI.
Reactive oxygen species, which in large measure are derived from PMNs, are one of the earliest and most important components of tissue injury after reperfusion of ischemic organs.18 Both BV, which is a direct product of HO-1 action, and BR, which is derived from BV, are potent physiological bile pigments with antioxidant function that protect cells from oxidative injury.19, 20 Presumably, the antioxidant potential of BV and BR are the mechanism in the models we have studied for the protection afforded by BV. Because oxidative stress plays a critical role in the development of IRI, BV can exert an important cytoprotective function, as in our study. The rationale for using BV instead of BR can be explained by its lower toxicity and the induction of BV reductase that can transform BV to BR, then amplifying its antioxidant action. Although BV is converted to BR by BV reductase, the BR can be recycled as BV, providing a powerful redox cycle.19 Moreover, the activity of BV against NO has been shown to be higher in comparison with BR.21 Recently it has been shown that BV therapy blocked hepatic acetaminophen-induced injury in rats by protecting the liver against reactive oxygen and nitrogen intermediates.22 Kato et al.23 have shown that BR rinse of livers decreased the oxidative stress and hepatobiliary dysfunction, mimicking the effects of HO-1–mediated preconditioning. In this report, BR used to rinse the grafts at a dose of 50 μmol/L resulted in a mortality rate of 100%.23 The same dose of BV in our study was nontoxic and resulted in 90% to 100% OLT survival. These combined results suggest that BV, which may have never converted into BR, acts as an important nontoxic cytoprotectant when used at the time of reperfusion, the period when oxygen-free radical–induced injury starts. In our model, BV donor treatment alone given 24 hours before injury was ineffectual. We believe this lack of effect is that BV is no longer present 24 hours after administration, and its effects have worn off.24
BV treatment effectively reduced the expression of TNF-α and IL-1, two cytokines implicated in hepatic IRI. Both cytokines upregulate the expression of adhesion molecules favoring leukocyte–sinusoidal endothelial cells (EC) interactions that result in additional cytokine production.25 Moreover, TNF-α induces local generation of a chemokine–epithelial neutrophil activating protein-78, which plays an important part in PMN chemotaxis/activation and induces Kupffer cells to generate superoxide radicals.26 IL-1 induces Kupffer cells to produce TNF-α and also upregulates free radical production. The role of both cytokines is further confirmed by experiments in which their neutralization decreased severity of reperfusion injury, as evidenced by decreased PMN infiltration and reduced damage to the parenchyma,27 the same effects we have achieved with BV therapy. Recently, a putative mechanism through which IL-1β is able to stimulate iNOS expression in hepatocytes has been proposed.28 The effect of BV therapy downregulating IL-1β expression can be another mechanism that explains the profound iNOS decrease in BV-treated OLTs. There are at least two reasons why the cytokines may be suppressed. Firstly, BV directly impacts the production of the cytokines. Secondly, BV leads to suppression of other factors that result in hepatic damage, and without that damage there is no stimulus for cytokine production.
HO-1 upregulation inhibits inflammatory responses, consistent with our immunohistochemical findings of markedly decreased mononuclear cell infiltration in BV-treated OLTs compared with untreated controls. Moreover, BV inhibited surface expression of P-selectin/E-selectin as well as ICAM-1. Whether this suppressive effect is a direct action of BV on the EC (as recently shown by us29) or the reflection of the decreased levels of cytokines that normally act on EC to induce adhesion molecule expression is not clear. The suppression is nevertheless important, because it may have inhibited the influx of PMNs and monocytes from the circulation into the liver during reperfusion. In fact, initial tethering of PMNs in sinusoidal venules requires expression of selectins on EC and interaction with their counter-receptors on PMNs. In venular endothelial cells, P-selectin/E-selectin as well as ICAM-1/ vascular cell adhesion molecule 1 - can be transcriptionally upregulated.30 BV reduces lipopolysaccharide-induced P-selectin/E-selectin expression in different vascular beds and especially in the liver that responds more dramatically to BV.31 In our OLT model, BV treatment reduced the expression of ICAM-1, which can represent an important component of its cytoprotective activity against IRI. Because increased ICAM-1 expression can be induced by various proinflammatory cytokines such as TNF-α and IL-1,32 it is not surprising that IRI resulted in increased ICAM-1 expression in the controls compared with the BV group. Consistent with these findings, blocking P-selectin–PMN interactions with anti–ICAM-1 antibodies or P-selectin glycoprotein ligand–immunoglobulin fusion protein (rPSGL-Ig) have been used effectively to attenuate liver injury in animal transplant models.33, 34
HO-1 overexpression exerts prominent hepatic cytoprotection against IRI in association with the modulation of pro- and antiapoptotic pathways.35 In this study, prolonged OLT survival was accompanied by enhanced local expression of Bcl-2/Bag-1 and decreased expression of proapoptotic caspase 3, which is consistent with our previous data.16 We have demonstrated that after HO-1 preconditioning, BV can inhibit most of the apoptosis seen. Interestingly, increased expression of antiapoptotic molecules promoted by BV was independent of HO-1, suggesting that BV may be the main mediator through which HO-1 prevents cell death. Our results are in agreement with in vitro observations indicating that BV-reductase depleted cells are more susceptible to caspase-dependent death from hyperoxia and to hydrogen peroxide toxicity.19 However, it has also been postulated that carbon monoxide may mediate the antiapoptotic effects of HO-1.16, 36
In conclusion, this study shows that BV, used as an adjuvant treatment during reperfusion, attenuates IRI in rat models of prolonged cold ischemia followed by ex vivo reperfusion or syngeneic transplantation. BV treatment of the host after OLT significantly inhibited liver PMN/macrophage infiltration through the inhibition of cellular adhesion molecules (P-selectin, ICAM-1), decreased the expression of proinflammatory cytokines (TNF-α, IL-1β), suppressed local expression of iNOS, and modulated pro- and antiapoptotic pathways independently of HO-1 expression. This study provides a rationale for a novel therapeutic approach using exogenous BV, a nontoxic product of the anti-inflammatory HO-1 pathway, to maximize organ function after transplantation and thus more effectively use the organ donor pool despite prolonged periods of cold ischemia.