We previously demonstrated in in vitro and ex vivo models that physiological concentrations of unconjugated bilirubin (BR) prevent oxidative stress (OS)-induced hepatocanalicular dysfunction and cholestasis. Here, we aimed to ascertain, in the whole rat, whether a similar cholestatic OS injury can be counteracted by heme oxygenase-1 (HO-1) induction that consequently elevates endogenous BR levels. This was achieved through the administration of hemin, an inducer of HO-1, the rate-limiting step in BR generation. We found that BR peaked between 6 and 8 h after hemin administration. During this time period, HO-1 induction fully prevented the pro-oxidant tert-butylhydroperoxide (tBuOOH)-induced drop in bile flow, and in the biliary excretion of bile salts and glutathione, the two main driving forces of bile flow; this was associated with preservation of the membrane localization of their respective canalicular transporters, bile salt export pump (Bsep) and multidrug resistance-associated protein 2 (Mrp2), which are otherwise endocytosed by OS. HO-1 induction counteracted the oxidation of intracellular proteins and membrane lipids induced by tBuOOH, and fully prevented the increase in the oxidized-to-total glutathione (GSHt) ratio, a sensitive parameter of hepatocellular OS. Compensatory elevations of the activity of the antioxidant enzymes catalase (CAT) and superoxide dismutase (SOD) were also prevented. We conclude that in vivo HO-1 induction protects the liver from acute oxidative injury, thus preventing consequent cholestasis. This reveals an important role for the induction of HO-1 and the consequently elevated levels of BR in preserving biliary secretory function under OS conditions, thus representing a novel therapeutic tool to limit the cholestatic injury that bears an oxidative background.
In aerobic organisms, the use of oxygen in the mitochondria as the last acceptor of the respiratory electron chain constitutes an evolutionary advantage. However, the main drawback is its high cytotoxic potentiality . Cellular oxygen is used in 97–98% for water synthesis and the other 2–3% continuously generates reactive oxygen species (ROS) that can damage cellular structures. Oxidative stress (OS) arises from the imbalance between ROS generation during metabolic processes and the capacity of antioxidant systems to clear them out . OS is a common feature in most hepatopathies, many of which concur with either mild or severe cholestasis. Obstructive cholestasis , sepsis-induced cholestasis , viral , toxic , and autoimmune hepatitis , alcoholic steatohepatitis , nonalcoholic fatty liver disease , and hepatic ischemia–reperfusion injury [9,10] are amongst those clinical entities bearing an oxidative background. Cumulative evidence indicates that ROS induce a number of functional changes either deleterious or adaptive in the capability of the hepatocytes to produce bile and to secrete exogenous and endogenous compounds. Functional changes involve impairment of biliary secretion through both, direct oxidative damage to the cellular structures involved in this process, and modification of intracellular redox-sensitive signal transduction pathways . OS-induced impairment of bile flow generation has been demonstrated to occur soon after exposure to a number of pro-oxidant agents, including tert-butylhydroperoxide (tBuOOH) [12,13]. This compound is a synthetic analog of short-chain lipid hydroperoxides formed endogenously under OS conditions, which has been widely used as a model to study the effect of OS on biological systems [14–16]. Apart from inducing OS directly via reduction into both peroxyl- and alkoxyl-free radicals by both cytochrome P450 and the mitochondrial electron chain , tBuOOH promotes OS via ROS-induced formation of the mitochondrial permeability transition pore (MPTP) and further mitochondrial production of ROS . Therefore, when administered acutely and studied in a short-term time scale, tBuOOH represents an ideal tool for the study of the effect of a ‘pure’ oxidant insult on liver secretory function and its prevention.
In previous work, our group demonstrated that endocytic internalization of canalicular transporters is a key factor in OS-induced cholestasis. We found in isolated rat hepatocyte couplets that tBuOOH induces endocytic internalization of the bile salt export pump (Bsep, Abcb11), a main canalicular carrier that transports into the bile canaliculus osmotically active monoanionic bile salts, the main driving forces involved in the generation of the so-called ‘bile salt-dependent bile flow’ . The same holds true for the multidrug resistance associated protein 2 (Mrp2, Abcc2), as has been shown in isolated rat hepatocyte couplets [20,21], and perfused rat livers exposed to the pro-oxidant agent tBuOOH [20,21]; this transporter mediates the biliary excretion of glutathione, the main driving force of the so called ‘bile salt-independent bile flow’ .
Bilirubin (BR), one of the end products of heme catabolic pathway, has been historically considered not only a waste product but also a potentially toxic compound. However, this view began to change by the late 20th century, owing to the work of Stocker and colleagues [23,24] and today BR is regarded as a potent endogenous antioxidant agent with cytoprotective properties [25,26]. Antioxidant effects of BR in vitro comprise the scavenging of singlet oxygen  and superoxide anion , as well as peroxyl radicals , thus explaining its protective effects against membrane lipid peroxidation . In addition, BR is a well-known scavenger of reactive nitrogen species . In vivo, BR has been found to protect rodents against both OS-induced retinal degeneration  and diabetic nephropathy , and it also shows anti-genotoxic effects in humans and animal models . BR has also been shown to bear an immunomodulatory function, as it prevents complement system-mediated hepatocyte lysis [34–36] and suppresses innate immune response after islet transplant . In our previous work, we demonstrated that physiological concentrations of unconjugated BR prevent OS-induced hepatocellular damage, and that this cytoprotective effect is exerted through prevention of actin cytoskeleton disarrangement and the consequent endocytic internalization of canalicular transporters crucial to normal biliary secretory function . These findings support the notion that unconjugated BR, an endogenously generated metabolite, may act as an efficient antioxidant agent, even at physiological concentrations.
Until nowadays, a myriad of studies has demonstrated the potential utility of exogenously administered BR in the treatment of several diseases, such as post-transplantation ischemia–reperfusion , cardiovascular , pulmonary , and metabolic diseases . However, studies inquiring into the effect of endogenously elevated levels of BR occurring during pathological conditions of the liver are scarce. Moreover, the likely beneficial action of heme oxygenase-1 (HO-1) induction and consequently elevated BR levels as a therapeutic strategy under a ‘pure’ OS-induced cholestatic condition has never been addressed. The present work was aimed to extend our previous in vitro/ex vivo findings to the whole animal, an in vivo model much closer to the human pathophysiological context, and to use an experimental therapeutic approach safe and feasible enough to be translated into clinical use for the controlled elevation of the endogenous levels of BR, based upon induction of HO-1, the rate-limiting step in the production of BR from heme degradation. This NADPH-dependent enzyme opens the porphyrin ring via oxidation of the heme α-carbon bridge, with the release of Fe2+ and the formation of carbon monoxide (CO) and biliverdin; this latter compound is rapidly converted into BR by biliverdin reductase . By using this approach, here we show for the first time that the induction of HO-1 at a level enough to significantly elevate endogenous BR levels is able to preserve the functional status of hepatocytes in terms of biliary secretory function under ‘pure’ OS conditions, acting as a crucial antioxidant defense to keep the secretory machinery operative after an oxidative challenge.
Materials and methods
Hemin, tBuOOH, thiobarbituric acid (TBA), 2,4-dinitrophenyhydrazine (DNPH), glutathione reductase from Saccharomyces cerevisiae, β-NAD 2′-phosphate reduced tetrasodium salt hydrate (β-NADPH), and 3α-hydroxysteroid dehydrogenase from Pseudomonas testosteroni were purchased from Sigma Chemical Co. (St. Louis, MO, U.S.A.). Rabbit anti-rat Bsep was from Kamiya Biomedical Co. (Seattle, WA, U.S.A.) and mouse anti-human MRP2 (M2III-6) was from Alexis Biochemicals (San Diego, CA, U.S.A.). Mouse anti-occludin, rabbit anti-occludin, Cy2–conjugated goat anti-mouse IgG, Cy2–conjugated donkey anti-rabbit IgG, and horseradish peroxidase (HRP)–conjugated goat anti-mouse IgG were from Thermo Fisher Scientific, Inc. (Waltham, MA, U.S.A.). Cy3–conjugated goat anti-rabbit IgG and Cy3–conjugated donkey anti-mouse IgG were obtained from Jackson ImmunoResearch Laboratory (West Grove, PA, U.S.A.). Mouse monoclonal antibody to HO-1 was from Abcam (Cambridge, U.K.). All the other chemicals were of the highest purity commercially available.
Animals and treatments
Animals were obtained from the Institute of Veterinary Sciences Litoral (ICiVet-Litoral, UNL - CONICET) and kept under identical conditions until use. All animals received humane care according to the criteria outlined in the ‘Guide for the Care and Use of Laboratory Animals’, written by the National Academy of Sciences and published by the National Institutes of Health (NIH publication 25-28, revised 1996) and the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines (http://www.nc3rs.org/ARRIVE). All experimental protocols were approved by the Bioethical Committee for the Care and Use of Laboratory Animals of the Faculty of Biochemical and Pharmaceutical Sciences of the National University of Rosario (Resolution N° 489/2015). Studies were performed on male Wistar rats (90 days) weighing 300–350 g, maintained on a standard diet and water ad libitum, and housed in a temperature- and humidity-controlled room under a constant 12-h light, 12-h dark cycle. During the experiments, animal welfare was warranted by monitoring vital signs (heart and respiratory rates) and pain sensitivity (paw-withdrawal latency). At the end of the experiments, animals were killed by overdose of anesthesia, followed by exsanguination.
In studies aimed to assess the capability of hemin to induce HO-1, animals were randomly divided into two experimental groups: (i) Hemin group (n=10), which was administered hemin (20 mg/kg b.w., i.p., dissolved in DMSO plus saline); (ii) Control group (n=10), which received only the hemin vehicle.
In studies aimed to assess the capability of hemin to prevent tBuOOH-induced OS and hepatobiliary dysfunction, animals were randomly divided into four experimental groups: (i) ‘Hemin+tBuOOH’ group (n=12 for each experiment), which was administered hemin (20 mg/kg b.w., i.p., dissolved in DMSO plus saline), and 2 h later, tBuOOH (440 µmol/kg b.w., i.p., dissolved in saline); (ii) ‘Hemin’ group (n=12 for each experiment), which received hemin, followed 2 h later, by the i.p. administration of the tBuOOH vehicle; (iii) ‘tBuOOH’ group (n=12 for each experiment), which was administered the hemin vehicle, followed 2 h later, by tBuOOH administration; (iv) ‘Control’ group (n=12 for each experiment), which received only the hemin and tBuOOH vehicles (Figure 1).
Timeline of experimental procedures
All experiments started at 7 a.m. each day, with the corresponding injection according to the day protocol. Surgical procedures began 4 h after tBuOOH/vehicle injection. For this purpose, animals were anesthetized using ketamine/xylazine (100 mg/3 mg/kg i.p.) and bile duct was cannulated by means of a PE-10 catheter. Bile samples were collected every 10 min, for 2 h. At the end of the collection period, animals were killed, and liver was removed, washed, and stored according to the protocols of the different assays carried out afterward, i.e. part of the organ was frozen in isopentane, and stored at −70°C for immunofluorescence staining, while another portion was homogenized in 0.3 M sucrose plus protease inhibitors, for the remaining studies.
Evaluation of the impact of hemin treatment on HO-1 expression and endogenous BR levels
Hemin (ferriprotoporphyrin IX), a well-known inducer of HO-1 , was used throughout. Different doses of hemin were tested, as well as different time points after hemin injection at the selected dose. Biliary concentration of BR was used as a surrogate parameter to estimate BR endogenous levels. For this purpose, BR was assessed by spectrophotometric determination of absorbance at 462 nm of the bile samples collected; since the absorption coefficients of BR esters are similar to each other, the obtained absorbance value results from the sum of the absorbances of each of the esters .
Western blotting studies were carried out in hepatic homogenates, in order to confirm HO-1 induction by hemin. Briefly, liver tissue was homogenized in 0.3 M sucrose plus protease inhibitor cocktail (Sigma Chemical Co., St. Louis, MO, U.S.A.), and disrupted by sonication. Aliquots of equivalent protein content, as ascertained by the Lowry method , were separated by electrophoresis on an SDS/12% polyacrylamide gel and electrotransferred to PVDF membranes. These membranes were then incubated overnight with a mouse monoclonal antibody to HO-1 (1:1000), and HRP–conjugated goat anti-mouse IgG (1:3000, 1 h) was used as the secondary antibody. Blots were also probed for β-actin (Sigma–Aldrich Corp.) as a loading control. Protein bands were visualized using a chemiluminescence reagent (ECL™-GE Healthcare, Amersham, U.S.A.) and captured with Hyperfilm ECL™ (GE Healthcare, Amersham, U.S.A.). For quantitation, blots from at least four independent experiments were quantitated using ImageJ 1.51p software (NIH, U.S.A.).
Evaluation of the cellular redox status in OS-induced acute cholestasis
The magnitude of the oxidative hepatic impact induced by tBuOOH was evaluated as follows: (i) by measuring the generation of TBA-reactive substances (TBARS) in liver homogenates ; (ii) by measuring the concentration of protein carbonyls in liver homogenates by the method described by Levine and colleagues , which is based on the reaction of carbonyl groups with DNPH to form the corresponding hydrazone, which is spectrophotometrically detected at 370 nm; (iii) by determination of the ratio of oxidized glutathione (GSSG)-to-total glutathione (GSHt) in bile, for which bile samples were subjected to assessment of both GSSG and GSHt, by the recycling method of Tietze , as modified by Griffith ; (iv) by assessment of the activity of antioxidant enzymes catalase (CAT) and superoxide dismutase (SOD). The activity of CAT was assessed by recording the decomposition of H2O2 at 240 nm . The activity of SOD was assessed by a commercial kit (Randox Ltd., Crumlin, U.K.), based on the utilization of xanthine and xanthine oxidase in order to generate superoxide anion, which then reacts with 5-phenyltetrazolium 2-(4-iodophenyl)-3-(4-nitrophenol)-chloride, forming a red complex, which is detectable at 420 nm.
On the other hand, presence of inflammatory cell infiltrate and levels of interleukin-6 were determined in liver slices and plasma samples respectively, of each studied group in order to discard any inflammatory changes occurring after tBuOOH treatment. Liver slices were stained with Hematoxylin–Eosin and interleukin-6 was assessed by means of a commercial ELISA kit provided by Thermo Fisher Scientific, Inc. (Waltham, MA, U.S.A.). Besides, in order to rule out steatosis as a component of our acute oxidative injury model, liver slices were stained with Sudan Black B and plasma samples were assayed for triglycerides and total cholesterol levels (CM250 Autoanalyzer, Wiener Lab, Rosario, Argentina).
Evaluation of biliary secretory function and canalicular transporter localization in OS-induced acute cholestasis
The impact of OS on normal secretory function was evaluated by studying the following parameters: (i) bile flow, gravimetrically assessed assuming a bile density of 1 g/ml; (ii) biliary excretion of bile salts, measured enzymatically by using 3α-hydroxysteroid dehydrogenase, as described by Talalay ; this parameter allows the estimation of the function of Bsep; (iii) biliary GSHt and BR concentrations, as described above, which estimate Mrp2 function; (iv) cellular localization of Bsep and Mrp2, by confocal immunofluorescent microscopy (LSM880, Carl Zeiss LLC, Thornwood, NY, U.S.A.). For Bsep and Mrp2 labeling, tissue sections were incubated overnight with the specific antibody to Bsep (1:100) or Mrp2 (1:100). Sections were then incubated either with Cy2–conjugated donkey anti-rabbit IgG (for the case of Bsep) or with Cy2–conjugated goat anti-mouse IgG (for the case of Mrp2). Immunolabeling of occludin was simultaneously processed (co-staining), so as to delimitate the canalicular region. For this purpose, mouse (for Bsep) or rabbit (for Mrp2) anti-occludin was used as primary antibody, followed by incubation with either Cy3–conjugated donkey anti-mouse IgG or Cy3–conjugated goat anti-rabbit IgG, respectively, for 1 h. Once stained, the specimens for Bsep and Mrp2 analysis were examined by confocal microscopy, and the images thus obtained were analyzed to quantitate the total intensity of fluorescence within the canalicular region, delimitated by occluding staining.
Results were expressed as mean ± S.D., with n representing the number of replicates of each experiment. The Student’s unpaired t test was used for comparison between two groups, as long as the requirements for parametric analysis were fulfilled; comparisons that did not meet these criteria were carried out using the Mann–Whitney’s rank sum test. When comparing more than two groups, the Kruskal–Wallis’ test (one-way ANOVA on ranks) was used; this was followed by the Dunn’s multiple-comparison, post hoc test for pairwise comparisons, if ANOVA reached any statistical significance amongst groups. Comparison of the variances of the densitometric profiles of Bsep and Mrp2 localization was performed using the Mann–Whitney U test. P-values <0.05 were considered to be statistically significant.
Effect of hemin treatment on HO-1 expression and endogenous BR levels
Hemin was used as an inducer of HO-1, and biliary levels of BR increased consequently. Hemin doses of 5, 10, and 20 mg/kg b.w. were tested and the biliary BR concentration measured 6 h post-hemin injection was significantly elevated at the dose of 20 mg/kg b.w. (Figure 2A), which was then selected to perform a time-course study of hemin effect. Animals injected with 20 mg/kg b.w. hemin were subjected to surgery at 2, 4, 6, 8, 12, and 24 h after hemin treatment, and biliary concentration of BR was assessed; values peaked at 6–8 h post-hemin injection, reaching approximately three-fold increase in its basal biliary levels (Figure 2B). From these experiments, a hemin dose of 20 mg/kg b.w. and a time point of 6–8 h post-hemin injection were selected for further studies. Plasma levels of BR did not increase at any dose nor at any time point after hemin injection (data not shown); this may be due to the extremely efficient hepatic detoxification of BR in the rat, with maximal phase II metabolic rate and biliary transport, amongst many species tested . Western blotting analyses of HO-1 protein expression corroborated the induction of HO-1 expression by hemin treatment (Figure 2C).
Effect of hemin treatment on HO-1 expression and endogenous BR levels
Effect of HO-1 induction on cellular redox status in acute oxidative cholestasis
Oxidative injury was evaluated through the ability of ROS to directly damage cellular macromolecules. For this purpose, hepatic lipid peroxidation and protein carbonylation were assessed. Animals treated with tBuOOH showed higher levels of TBARS and carbonylated proteins as compared with the Control group, while in animals pre-treated with hemin, similar values to Controls were obtained (Figure 3). GSSG/GSHt relation in bile, which is considered as a sensitive marker of hepatic OS , was also measured. We found a three-fold higher GSSG/GSHt ratio in animals subjected to the oxidative insult, while pre-treatment with hemin attenuated this rise (Figure 4). Altogether, these results clearly indicate a protective effect exerted by HO-1 induction against the oxidative injury induced by tBuOOH, likely due to increased levels of BR.
HO-1 induction prevents oxidative damage to lipids and proteins
HO-1 induction prevents the increase in GSSG/GSHt ratio in bile
In order to evaluate the status of antioxidant defenses, levels of CAT and SOD were determined in liver tissue homogenates. We found that hepatic levels of both antioxidant enzymes were increased by tBuOOH, and that this elevation was abolished by hemin pre-treatment. Superoxide anion radical metabolization by SOD converts it into hydrogen peroxide, which is then converted into water and oxygen by CAT. Therefore, our results show that HO-1 induction prevents the ROS-mediated triggering of the antioxidant defense machinery in the liver, an effect most likely due to the scavenging properties of BR, which prevents ROS from building up (Figure 5). Hemin treatment alone produced no significant variation in any of the parameters studied.
HO-1 induction prevents the triggering of the antioxidant response
At the same time, we corroborated that tBuOOH-induced injury is of strictly pro-oxidant nature, since no evidence of either inflammation (e.g. leukocytes infiltration and interleukin-6 elevations) or steatosis (e.g. hepatic lipid droplets and increased triglycerides and total cholesterol levels) was observed (Figure 6).
tBuOOH induces a ‘pure’ oxidative injury
Effect of HO-1 induction on biliary secretory function in acute oxidative cholestasis
Once evidenced the oxidative injury elicited by tBuOOH, we evaluated whether the main parameters of biliary secretory function were consequently altered. Bile flow was significantly decreased in animals treated with tBuOOH throughout the period of time studied, while pre-treatment with hemin completely prevented this drop (Figure 7). In order to investigate whether prevention exerted by pre-treatment with hemin on bile flow drop was due to its protective effect on the secretory capacity of key canalicular transporters, we evaluated the biliary excretion of Bsep and Mrp2 specific substrates. As expected, biliary excretion of bile salts decreased in animals treated with tBuOOH, and pre-treatment with hemin prevented this decrease, thus evidencing a protective effect of HO-1 induction on Bsep function (Figure 8A). A similar phenomenon was observed for Mrp2. Biliary excretion of the Mrp2 substrates, BR and GSHt, dropped after tBuOOH treatment, while hemin prevented such a decrease (Figure 8B,C). Taken together, these results show that HO-1 induction, most likely through elevation of endogenous BR levels, prevents the impairment in the function of the two main canalicular transporters involved in bile formation, thus preventing OS-induced acute cholestasis.
HO-1 induction prevents OS-induced bile flow impairment
HO-1 induction prevents the oxidative impairment in biliary secretory function
Since proper plasma membrane localization is crucial to Bsep and Mrp2 transport function, and OS alters their stability in membrane leading to endocytic internalization, we ascertained whether HO-1 induction with hemin can prevent this alteration, by fluorescent immunostaining of these transporters, followed by confocal microscopy. For this purpose, tissue sections were subjected to specific co-staining of Bsep or Mrp2 with occludin. In control livers, transporter-associated fluorescence was confined to the canalicular space, whereas tBuOOH induced redistribution of both transporters from the canalicular membrane into vesicles located either in cytosol or around the pericanalicular area. Pre-treatment with hemin significantly prevented these alterations, as can be seen both in confocal images (Figure 9) and through the quantitative analysis of the images obtained (Supplementary Figure S1). These results strongly suggest that the protective action of HO-1 induction, and consequently elevated levels of BR, on the oxidative impairment of biliary secretory function is also exerted at the level of canalicular localization of transporters that are crucial to the biliary secretion process.
Endocytic internalization of Bsep and Mrp2 induced by OS, and its prevention by HO-1 induction
In the present study we show, for the first time, that induction of HO-1 by hemin protects the liver from a ‘pure’ oxidative cholestatic injury in vivo, most probably due to the consequently elevated levels of BR. Although it is a well-known antioxidant metabolite [53,54], BR has scarcely been regarded as a therapeutic strategy for hepatopathies. This may certainly be due to its also well-recognized deleterious effects when present at high concentrations. These effects have been described to mainly affect the nervous system [55,56]. Nevertheless, under controlled conditions, BR exhibits hepatoprotective properties from what to take benefit in the case of exogenously administered inducers of its synthesis.
In a previous work frorm our group, we found that physiological concentrations of unconjugated BR prevent the biliary secretory function to collapse under an OS insult in in vitro models, such as isolated rat hepatocytes, isolated rat hepatocyte couplets, and isolated and perfused rat liver . In light of the potential use of this endogenous metabolite for the treatment of acute oxidative cholestasis, we tested the effect of HO-1 induction, a key step in BR formation, after in vivo exposure to mild doses of tBuOOH, a well-known model pro-oxidant agent. tBuOOH, administered i.p. to adult rats on a specific time point, and at a dose of 440 µmol/kg b.w., caused significant OS, as evidenced by the elevated tissular lipid and protein oxidation levels, the increased biliary GSSG/GSHt ratio, and the enhanced antioxidant compensatory response in liver tissue.
Here, we found that ROS generated by tBuOOH induce peroxidation of membrane lipids and carbonylation of cellular proteins. Lipid peroxidation is initiated when a hydrogen atom is extracted from an unsaturated fatty acid by a radical species, thus triggering a destructive chain reaction that generates lipid peroxides. Disruption of cellular membranes by these peroxides produces reactive metabolites like malondialdehyde and 4-hydroxy-2,3-transnonenal, causing cellular dysfunction . On the other hand, carbonyl groups (aldehydes and ketones) are introduced into proteins by primary oxidative modifications, and the appearance of such carbonyl groups is taken as a likely evidence of oxidative modification . Induction of HO-1 prevented such oxidative deleterious events most likely by the scavenging effect of BR on the radical species generated after tBuOOH injection. In this sense, many authors have demonstrated that BR is endowed with a strong antioxidant activity [25–30]. Amongst the most recent ones, Zelenka and colleagues  described that BR induces a dose-dependent decrease in cytoplasmic and mitochondrial levels of hydrogen peroxide in a human embryonic kidney cell line, and Jansen and colleagues  found that BR inhibits mitochondrial ROS formation, and that it is a potent scavenger of superoxide radical.
BR, at a concentration in the nM order, is able to protect against 10000-fold higher concentrations of H2O2 , since the antioxidant actions of BR may be amplified 10000 times or greater by the biliverdin reductase cycle . Although biliverdin is the actual product of HO-1 activity, and BR is produced after a further reduction step catalyzed by biliverdin reductase, the possibility that biliverdin, rather than BR, is involved in the antioxidant effects of HO-1 induction, needs to be considered. There is agreement in that biliverdin bears some ROS-scavenging properties in vitro , but its antioxidant activity is 60-fold lower than BR, because, unlike BR, it lacks a γ-methene group . In addition, biliverdin is immediately reduced to BR, so that its intracellular levels are expected to be low . Furthermore, studies with full HO-1 activity but inhibition of biliverdin reductase, ensuring biliverdin but not BR formation, seems to support this concept, since the antioxidant effect afforded by HO-1 induction was fully inhibited [62,63]. In addition, inhibition of biliverdin reductase in vivo exacerbates OS in diseases with an oxidant background, such as diabetes  and Alzheimer disease ; this finding was reproduced in biliverdin reductase-depleted cells, where ROS-mediated cytotoxicity was exacerbated. Interestingly, the pro-oxidant effect of biliverdin reductase depletion was even higher than that recorded after glutathione depletion, suggesting that BR exerts an even higher antioxidant effect than glutathione in the cell . Irrespective of their relative antioxidant potency, both compounds seem to have complementary antioxidant properties; unconjugated BR would protect preferentially membrane lipids and membrane-associated proteins due to its high hydrophobicity, whereas glutathione preferentially protects soluble proteins due to its hydrophilicity . Our finding that HO-1 induction fully protected against tBuOOH-induced lipid peroxidation further points BR rather than biliverdin as the actual antioxidant compound, since unlike BR, biliverdin is highly water soluble .
Another HO-1 product that may bear, at least hypothetically, protective effects against tBuOOH-induced OS is CO. tBuOOH induces OS, in part, by inducing MPTP formation, and further mitochondrial production of ROS , and CO was shown to limit this process by inducing a mild-uncoupling state in isolated mitochondria by modulating heme-containing cytochrome oxidases . However, tBuOOH only induced MPTP formation at a concentration of 500 µM , a level that was shown to virtually abolish canalicular secretory function in an isolated rat hepatocyte couplets model ; this is in contrast with our results here in the in vivo tBuOOH cholestatic model, where only a partial decrease in bile flow was recorded, suggesting that such a high tBuOOH level was not reached. Taken together, our own results and those available in the literature and summarized above rather point to endogenously produced BR as the major agent mediating HO-1 antioxidant effects.
Although abundant in in vitro experimental systems, compelling evidence in in vivo settings for an antioxidant cytoprotective role of BR is scarce. Boon and colleagues  demonstrated decreased circulating OS biomarker levels in Gunn rats, although they did not find alterations in cell-based OS markers, thus assuming that BR protective effects are confined to the vascular compartment. Contrarily, here we demonstrated a clear hepatocellular antioxidant protective effect of BR due to locally increased levels of this metabolite, while no elevations in systemic BR concentrations were found.
In the last years, evidence has accumulated that OS is cholestatic due to an impairment of hepatocellular biliary secretory machinery, mainly through direct oxidative damage to cellular structures by ROS . However, as the liver constitutes an organ that is particularly susceptible to OS due to its high metabolic activity, it is equipped with special defense mechanisms to scavenge ROS, with glutathione and the antioxidant enzymes CAT and SOD amongst the main ones. For this purpose, the liver has redox-sensitive mechanisms that lead to the induction by ROS of these and other antioxidant enzymes, including HO-1 itself, as a compensatory response aimed to decrease ROS generated by the oxidative insult and re-establish the redox balance. This OS-triggered induction is mainly mediated by the redox-sensitive transcription factor Nuclear factor-erythroid related factor-2 (Nrf2) . Under normal conditions, Nrf2 is bound in cytoplasm to its repressor, Kelch-like ECH-associated protein (Keap1), an OS sensor. Under OS conditions, Nrf2 is released and translocated to the nucleus, where it binds to the antioxidant response element (ARE) located at the promoter region of HO-1 and other antioxidant enzymes, such as SOD, CAT, glutathione S-transferase, and peroxidase, thus transcriptionally promoting their synthesis . We studied this antioxidant adaptive response by evaluating the capability of tBuOOH to activate CAT and SOD, and the prevention of this effect by HO-1 induction. Both CAT and SOD activities were found to peak in animals treated with tBuOOH and pre-treatment with hemin prevented the activation of these antioxidant mechanisms of defense, likely due to BR scavenger action that provides a more reductive milieu, thus favoring a rapid and efficient protection against ROS produced by tBuOOH. Although tBuOOH is expected to induce HO-1 as a part of the antioxidant response induced by OS , it is apparent that the extent of this adaptive induction was not enough to counteract tBuOOH-oxidizing damage, and required the therapeutic addition of a potent inducer like hemin to complement this protection. Additionally, hemin is not only an inducer but also a substrate of HO-1, thus guaranteeing that its induction results in elevated levels of BR, the product of the reaction.
It has been well established that the oxidative challenge affects the hepatocyte secretory machinery by impairing both bile flow and the biliary excretion of endo- and xenobiotics . Here, we demonstrated that the excretion of specific substrates of hepatocanalicular transporters crucial to bile formation, such as Bsep and Mrp2, are highly impaired due to the oxidative insult, thus producing a decrease in bile flow. Localization of Bsep and Mrp2 at the hepatocanalicular membrane is a key feature for the normal function of these transporters and OS has been demonstrated to induce their internalization to the pericanalicular area in in vitro studies . Here, we reproduced the findings mentioned above in the whole animal context, thus enabling us to corroborate the delocalization phenomenon as the main cause of cholestasis. Taken together, our results suggest that the induction of HO-1 would counteract the outcome of cholestatic conditions that concur with OS.
Bile salt-independent bile flow mainly depends on the canalicular transport of glutathione, an Mrp2 substrate . Glutathione biliary excretion rate was decreased by tBuOOH treatment, thus suggesting a functional alteration of this transporter. Our results strengthen the hypothesis that tBuOOH-associated OS would reduce Mrp2 functional capacity, thus resulting in decreased excretion of glutathione. Since conjugated BR is also a substrate of Mrp2 , this functional failure would also explain the decrease in the excretion of BR in animals exposed to tBuOOH. On the other hand, we confirmed that induction of HO-1 increased hepatic BR levels, since biliary excretion of BR was increased and that of glutathione was decreased in animals treated only with hemin. This indicates that elevated intrahepatic levels of BR competitively inhibited the transport of glutathione by Mrp2. For this reason, we observed a partial prevention of the decrease in the biliary excretion of glutathione and a complete prevention of the decrease in the biliary excretion of BR in animals pre-treated with hemin. Bile salts excretion rate was also decreased by OS and counteracted by hemin, thus evidencing a functional oxidative impairment of Bsep that could result from a competitive inhibition process exerted by GSSG, excessively generated under these conditions , together with the endocytic internalization of Bsep induced by tBuOOH [19–21], two phenomena that were prevented by HO-1 induction. Our results are partially in-line with those obtained by Donner and colleagues  in a model of cholestasis by hepatic ischemia–reperfusion, where cholestasis and internalization of Mrp2 and Bsep were partially prevented by HO-1 induction. However, unlike our ‘pure’ OS model based upon the acute administration of a direct pro-oxidizing agent like tBuOOH, hepatic ischemia–reperfusion has a complex pathogenesis, involving inflammatory and microcirculatory alterations, apart from oxidant ones, which may have significantly contributed to the cholestatic failure . Importantly, they can be counteracted by other HO-1 products, as for example, CO, which has potent anti-ischemic and anti-inflammatory properties . Nevertheless, there is considerable evidence implicating ROS as another cause of the hepatic injury, since ischemia results in a build-up of NADH that, on reperfusion, generates a burst of ROS from the mitochondrial electron transport chain, which can be protected by HO-1-derived BR in the same manner it afforded full protection in our pro-oxidant model .
In light of the crucial role of OS in hepatopathies, particularly in the cholestatic ones involving malabsorption of fat-soluble antioxidant vitamins , antioxidants are considered a valid therapeutic approach for their treatment. However, such antioxidant strategies are nowadays scarce and not completely free from deleterious side effects. This is particularly worrying in long-term treatments with lipophilic antioxidants, such as vitamins E and A, since it has been described that both increase the probability of developing different types of cancer , and increase mortality in those patients already suffering from them . Taken together, our results allow us to postulate that an increase in the endogenous generation of BR by HO-1 induction could represent a much more beneficial and innocuous therapeutic strategy for chronic cholestatic diseases, such as primary sclerosing cholangitis or primary biliary cirrhosis, in order to minimize the deleterious effects associated with OS in such pathologies . In this sense, a decrease in HO-1 levels has been recently described in human cholestatic diseases such as primary biliary cirrhosis  and primary sclerosing cholangitis . In addition, intrahepatic bile salts accumulated in cholestasis, apart from being pro-oxidants themselves, also exacerbate OS by depleting hepatocytes of BR via induction of BR basolateral export pumps (e.g. Mrp3), and by exacerbating BR consumption via BR oxidation . Therefore, inducing this enzyme would clearly result in a beneficial counteraction. Nevertheless, BR metabolism is somewhat different between rat and humans. For instance, unlike humans, rat lacks gallbladder , thus resulting in a more diluted bile, since the main function of the gallbladder is to concentrate the bile before it is released into the duodenum. Also, rat bears lower plasma levels of BR , as compared with humans, due to an extremely efficient detoxification system. These facts should be borne in mind at the time of designing an HO-1 induction protocol in humans, since the impact of hemin treatment and consequent HO-1 induction is expected to be quantitatively different in terms of BR production. Then, doses and timings should be adjusted in order to achieve supraphysiological, but not harmful, levels of BR, such as those obtained here for a hemin dose of 20 mg/kg b.w. and at a time-dependent response of 6–8 h post-treatment. All in all, our results strongly encourage the development of new strategies aiming at increasing the endogenous bioavailability of BR by means of the strictly controlled manipulation of HO-1 expression.
Several studies have demonstrated the potential utility of exogenously administered BR in the treatment of various diseases. However, the likely beneficial action of HO-1 induction and consequently endogenously elevated BR as a therapeutic strategy under a ‘pure’ OS-induced cholestasis has never been addressed.
HO-1 induction and consequent elevation of BR fully prevented the OS-induced drop in bile flow and in the biliary excretion of bile salts and glutathione, two main driving forces of bile flow; this was associated with preservation of the localization of their canalicular transporters, otherwise endocytosed by OS. HO-1 induction counteracted the oxidation of proteins and lipids and the increase in the oxidized-to-GSHt ratio in bile, a sensitive marker of OS. Compensatory elevations of antioxidant enzymes were also prevented.
By extending our previous in vitro/ex vivo findings to the whole animal, an in vivo model much closer to the human pathophysiological context, the present study strongly encourages the development of new strategies aiming at the strictly controlled induction of HO-1 expression which renders elevated levels of the antioxidant metabolite BR.
We thank Dr Rodrigo Vena and Dr Verónica Livore for their valuable technical assistance in confocal imaging and protein carbonyls assay procedure, respectively.
P.L.M. and P.C. carried out all the experiments, equally contributing to the work. M.V.R. assisted in animal management and surgical procedures as well as in Western blotting technique. G.B.P. and A.I.M. performed all histological studies. D.E.A.F., S.M.M.A., and E.J.S.P. contributed to the interpretation of the results. C.L.B. wrote the manuscript with support from P.L.M., P.C., S.M.M.A., E.J.S.P., and M.G.R. C.L.B. designed and directed the project, aided by M.G.R. and E.J.S.P. All authors discussed the results and contributed to the final manuscript.
The authors declare that there are no competing interests associated with the manuscript.
This work was supported by the Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) [grant numbers PICT 2015-1242, PICT 2016-1613]; and the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) [grant number PIP 2013-112 201201 00217 (to C.L.B. and M.G.R.)].
bile salt export pump
mitochondrial permeability transition pore
multidrug resistance-associated protein 2
nuclear factor-erythroid related factor-2
reactive oxygen species
These authors contributed equally to this work.