Potential conflict of interest: Nothing to report.
This work was supported by the Deutsche Forschungsgemeinschaft (SFB 841 project C2, to G.S. and G.T., and SFB 841 graduate school “Inflammation and Regeneration,” to G.T.).
Induction or overexpression of the heme-degrading enzyme, heme oxygenase 1 (HO-1), has been shown to protect mice from liver damage induced by acute inflammation. We have investigated the effects of HO-1 induction in a mouse model of chronic liver inflammation and fibrogenesis with progression to hepatocellular carcinoma (HCC) (Mdr2ko; FVB.129P2-Abcb4tm1Bor). HO-1 was induced in vivo by treatment with cobalt protoporphyrin IX, starting at week 5 or 12 of mice lifespan, and continued for 7 weeks. Our results showed that HO-1 induction reduced liver damage and chronic inflammation by regulating immune cell infiltration or proliferation as well as tumor necrosis factor receptor signaling. Fibrosis progression was significantly reduced by HO-1 induction in mice with mild, as well as established, portal and lobular fibrosis. HO-1 induction significantly suppressed hepatic stellate cell activation. During established fibrosis, HO-1 induction was able to revert portal inflammation and fibrosis below levels observed at the start of treatment. Moreover, hepatocellular proliferation and signs of dysplasia were decreased after HO-1 induction. Conclusion: Induction of HO-1 interferes with chronic inflammation and fibrogenesis and, in consequence, might delay progression to HCC. (HEPATOLOGY 2012;)
Heme oxygenase 1 (HO-1) plays an essential role in heme catabolism, where it catalyzes oxidative degradation of heme to carbon monoxide (CO), free iron, and biliverdin, which is subsequently converted to bilirubin by bilirubin reductase.1, 2 Deficiency in HO-1 results in detrimental inflammatory processes,3 whereas induction by, for example, cobalt protoporphyrin IX (CoPP) or elevated expression by genetic aberrations4 interfere with inflammation. HO-1 has been shown to decrease proinflammatory cytokines, such as tumor necrosis factor alpha (TNF-α), in vitro5, 6 as well as in vivo.7, 8 We could show previously that induction or overexpression of HO-1 reduces liver damage in mouse models of acute inflammation.7, 8 Here, we investigated the effects of HO-1 induction on chronic inflammation and fibrosis of the liver. Chronic hepatitis, induced by, for example, viral infections or chronic exposure to toxins represents a severe health problem, because it bears a risk of progression to hepatocellular carcinoma (HCC), which represents the fifth most common cancer worldwide.9, 10
To investigate HO-1 effects on chronic hepatitis, we used an animal model of chronic portal inflammation and bile duct proliferation with progression to liver fibrosis and HCC (multidrug resistance transporter 2 knockout [Mdr2ko] mouse; FVB.129P2-Abcb4tm1Bor).11 Mdr2ko mice are lacking the Mdr2 P-glycoprotein, a phosphatidylcholine transporter, resulting in dysfunctional phospholipid secretion.12 Signs of inflammation (e.g. intense hepatic leukocyte infiltrations), which appear within the first weeks of age, are accompanied by an increase in plasma transaminase levels and followed by enhanced connective tissue storage and progression to fibrosis.13 Fibrogenesis is a dynamic process associated with activation of hepatic stellate cells (HSCs), which is characterized by collagen and alpha smooth muscle actin (α-SMA) expression as well as chemokine secretion and activation of matrix metalloproteinases (MMPs) and their inhibitors, tissue inhibitors of matrix metalloproteinases (TIMPs).14 As a consequence of chronic inflammation and progressing fibrosis, Mdr2ko mice have been shown to develop HCC within 12-15 months of age.15
We investigated the effect of HO-1 induction on chronic inflammation and fibrogenesis in mice with mild or established fibrosis. Our results show that HO-1 induction decreased chronic inflammation by regulating immune cell infiltration or proliferation, TNF receptor (TNFR) expression, and subsequent events in proinflammatory cytokine signaling, such as extracellular signal-related kinase (ERK) phosphorylation. Liver damage was significantly ameliorated for at least 8 weeks beyond treatment. HO-1 induction improved lobular fibrosis as well as portal inflammation to a state observed before the onset of treatment. In conclusion, induction of HO-1 interfered with chronic inflammation and fibrosis formation.
Mdr2 knockout (Mdr2ko; FVB.129P2-Abcb4tm1Bor) mice were kindly provided by Frank Lammert (Homburg, Germany). For all experiments, female mice were used, which display a more severe pathology than male mice.16 All mice received human care according to the guidelines of the National Institutes of Health as well as to the legal requirements in Germany. They were maintained under controlled conditions (22°C, 55% humidity, and 12-hour day-night rhythm) and fed a standard laboratory chow.
Dosage and Application Routes.
CoPP (Frontier Scientific Europe, Ltd., Lancashire, UK) was administered to mice intraperitoneally at 5 mg/kg twice-weekly; CoPP was dissolved in NaOH (0.2 M), adjusted to pH 7.6 with HCl (0.1 M), and filled with water to a stock solution of 1 mg/mL. Control mice were treated equally using solvent. Treatment was started at 5 or 12 weeks of age and continued for 7 consecutive weeks. For long-term follow-up experiments, mice were treated for 9 consecutive weeks, also starting at week 5 of age. For in vitro experiments, CoPP was used at 10 μg/mL.
Determination of Plasma Transaminase and Cytokine Levels.
Liver damage was assessed by measuring plasma enzyme activity of alanine aminotransferase (ALT),17 using an automated procedure (COBAS MIRA; Roche, Basel, Switzerland). Enzyme-linked immunosorbent assays (TNF, TNFR1, and TNFR2) were performed as described previously.18
Cell Isolation and Fluorescence-Activated Cell Sorting.
Isolation of primary hepatocytes (PHs) as well as fluorescence-activated cell-sorting (FACS) analysis were performed as described previously.18 HSCs were isolated, as described previously,19 and activated with 2 ng of transforming growth factor beta 1 (TGF-β1)/mL (Chinese hamster ovary–derived; R&D Systems, Minneapolis, MN).
Detection of Messenger RNA by Real-Time Reverse-transcriptase Polymerase Chain Reaction.
Isolation of total RNA, complementary DNA synthesis, and real-time reverse-transcriptase polymerase chain reaction (RT-PCR) were carried out as described previously.20 Oligonucleotides for subsequent PCR reactions were obtained from Metabion International AG (Martinsried, Germany) and are summarized in Table 1.
Table 1. Oligonucleotide Sequences for Real-Time RT-PCR
Protein Isolation From Mouse Liver and Western Blotting Analysis.
Protein lysates were prepared as described previously.21 Semiquantitative evaluations were performed using the VersaDocM Imaging System 4000 MP (Bio-Rad, Munich, Germany) after quantification with Image Lab Software (Bio-Rad).
Liver tissue was fixed in 4% phosphate-buffered formalin and embedded into paraffin. Hematoxylin and eosin (H&E) stainings of 3-μm paraffin sections were used to evaluate basic histomorphology of the specimens, especially the portal and lobular amount of inflammatory cells (score 0-3 for negative to strong infiltration) and signs of regeneration/degeneration of hepatocytes (e.g., cytoplasmic volume, nuclear polymorphism, and thickness of liver cell plates as well as hepatic apoptotic bodies).23 Content of extracellular collagen was measured by the amount of chromotrope-aniline blue (CAB) in the portal tract and hepatic lobule (score 0-3 for negative to strong staining).24 Proliferating cell nuclear antigen (PCNA) and desmin staining were performed on an autostainer system (Dako), using the EnVision Detection System (Dako, Glostrup, Denmark), and developed using diaminobenzidine as the chromogen substrate (Roche Molecular Biochemicals, Mannheim, Germany), according to the manufacturer's instructions. To highlight neutrophil granulocytes in liver tissue, naphtol AS-D chloroacetate esterase enzyme (NASD) histochemistry was used.25 Liver slices were stained for cluster of differentiation (CD)3 and forkhead box protein 3 (Foxp3) and developed with the PolapKit (Zytochem, Las Condes, Chile). Images were evaluated per mm2 (PCNA) or high-power field (400× magnification) using ImageAccess Enterprise software (version 9; Imagic, Glattbrugg, Switzerland). Histological quantification was performed by a pathologist.
Results were analyzed using the Student's t test, if two groups were compared; in addition, for analysis of real-time RT-PCR, logarithmized data and Welch's t test were used. All data in this study are expressed as a mean ± standard error of the mean. P ≤ 0.05 was considered significant.
HO-1 Induction Interferes With Chronic Liver Inflammation.
HO-1 induction or overexpression have been shown to interfere with liver damage in mouse models of acute inflammation and apoptosis.7, 8 We have investigated the effects of HO-1 induction on chronic hepatic inflammation and fibrosis in Mdr2ko mice.
Nine weeks of HO-1 induction in Mdr2ko mice (weeks 5-14 of lifespan) decreased plasma ALT levels, which increased over the first 22 weeks of lifespan (Fig. 1A, open bars). Significant improvement was observed for at least 8 weeks beyond treatment (Fig. 1A, closed bars). HO-1 induction by CoPP in liver tissue was detectable on messenger RNA (mRNA) (Supporting Fig. 1A) and protein level (Supporting Fig. 1B) and did not alter hematocrit values (Supporting Fig. 1C). Likewise, isolated PHs and HSCs of Mdr2ko mice showed HO-1 induction after CoPP incubation (Supporting Fig. 1D). IHC staining of liver samples obtained at 12 weeks of age revealed reduced periportal and lobular inflammation in CoPP-treated mice (Fig. 1B). HO-1 induction significantly reduced the expression of TNFR1 and 2, whereas TNF expression in whole liver tissue was increased in Mdr2ko mice, compared to FVB background control (data not shown), but was not altered by HO-1 induction (Fig. 1C). Additionally, plasma levels of soluble TNF receptors were elevated by CoPP treatment, indicating higher shedding rates (Supporting Fig. 1E). Furthermore, downstream targets of TNF-signaling were found to be regulated. Although equal amounts of p38 protein were detected, phosphorylated p38 (pp38) was significantly reduced in CoPP-treated Mdr2ko mice (Fig. 1D; Supporting Fig. 1F). Similarly, protein levels of total Erk42 and phosphorylated Erk44 (pErk44) were significantly reduced in CoPP-treated Mdr2ko mice (Fig. 1D; Supporting Fig. 1F), affirming decreased proinflammatory signaling.
Expression of the immune cell attractant, osteopontin (OPN),26 was found to be significantly decreased upon HO-1 induction at 12 weeks (Fig. 1C) and also at 19 weeks of age (data not shown). In fact, cell counting revealed reduced total numbers of hepatic leukocytes after HO-1 induction, whereas total amounts of hepatic leukocytes were elevated in Mdr2ko mice, in comparison to FVB background mice (Supporting Fig. 2A). Staining liver slices of 12-week-old Mdr2ko mice for CD3+ or Foxp3+ cells revealed increased amounts of both cell types in Mdr2ko mice, whereas HO-1 induction decreased those cell counts (Fig. 2A-C). Quantification showed that in periportal (Fig. 2B), but not in lobular (Fig. 2C), tracts of CoPP-treated Mdr2ko mice, the ratio between CD3+ and Foxp3+ cells was shifted toward Foxp3+ cells (1:0.48 versus 1:0.63; Fig. 2B), indicating a higher immunosuppressive status in CoPP-treated animals. Additionally, the population of Gr1+CD11b+ cells (including Gr1high and Gr1intermediate cells) among all leukocytes revealed significant reduction by HO-1 induction (Supporting Fig. 2B, representative dot plots, and 2C, quantification). Further gating for Gr1 and CD11b demonstrated an overrepresentation of Gr1intCD11bhigh cells after HO-1 induction (Supporting Fig. 2B,D). Moreover, the population of neutrophil granulocytes (Gr1highCD11bhigh) was reduced in Mdr2ko mice upon HO-1 induction (54.9% ± 2% versus 63.3% ± 2.2%; Supporting Fig. 2B, upper gate), whereas the frequency of Gr1intCD11bhigh cells was increased in CoPP-treated Mdr2ko mice, compared to solvent-treated Mdr2ko mice (45.1% ± 2% versus 36.7% ± 2.2%; Supporting Fig. 2B, lower gate, and 2C). Because of the typical light-scatter characteristics of monocytes/macrophages (dark gray area in the dot plot of forward- versus side-scatter characteristics), this population of Gr1intCD11bhigh cells might represent a phenotype of monocytic myeloid-derived suppressor cells (mMDSCs) (Supporting Fig. 2B).
Similarly to CD3+ and Foxp3+ cells, histochemistry revealed significantly more macrophages (F4/80; Fig. 2D) as well as neutrophil granulocytes (NASD; Fig. 2E) in livers of Mdr2ko mice. HO-1 induction reduced periportal and lobular macrophages (Fig. 2D), as well as periportal neutrophil granulocytes (Fig. 2E), significantly.
HO-1 Induction Interferes With Progression of Fibrosis.
Livers of 12-week-old solvent- or CoPP-treated Mdr2ko mice were analyzed for fibrosis formation. CAB staining revealed beginning septa formation detectable in portal tracts and hepatic lobules of solvent-treated Mdr2ko mice (Fig. 3A), whereas in CoPP-treated mice, only mild fibrosis was observed, which was limited to the portal tracts (Fig. 3A). The equilibrium of assembly and disassembly of the extracellular matrix (ECM) is regulated by the activity of MMPs.27 We observed that activity of MMP-9 was significantly up-regulated in Mdr2ko mice upon HO-1 induction, whereas activity of MMP-2 was not altered (Fig. 3B,C). The fibrosis marker, hydroxyproline, was also found to be significantly reduced in livers of CoPP-treated Mdr2ko mice (Fig. 4A). Fibrogenesis is characterized by high expression levels of collagens I and III,28 which we found to be reduced upon HO-1 induction (Fig. 4B).
Because activated HSCs play a crucial role in fibrosis, we stained for the HSC marker, desmin, and found those cells increased in Mdr2ko mice, whereas HSCs were reduced after CoPP-treatment (Fig. 4C). Staining for α-SMA revealed a large number of activated HSCs in Mdr2ko mice, which was significantly reduced in livers of CoPP-treated animals (Fig. 4D). To specifically investigate the effect of HO-1 induction on HSC activation, we isolated HSCs from wild-type mice, activated those cells by incubation with TGF-β1 and induced HO-1 in vitro. Our results showed that HO-1 induction significantly reduced the expression of the HSC activation marker, α-SMA, TNFRs, as well as expression of TGF-β1 (Fig. 4E), whereas it enhanced the expression of HO-1, TIMP-1, MMP-9, and MMP-13 (Fig. 4F).
Established Fibrosis Is Reduced Upon HO-1 Induction.
To investigate the effects of HO-1 induction on established fibrosis, we started treatment of Mdr2ko mice at the age of 12 weeks. Measurement of plasma ALT levels after 7 weeks of CoPP treatment revealed that induction of HO-1 significantly decreased hepatocyte damage (Supporting Fig. 3A). Similar to our observations during early fibrosis (Fig. 4B), CoPP treatment decreased the expression of collagens I and III (Supporting Fig. 3B) and elevated MMP-9 activity significantly (data not shown). Inflammation (Supporting Fig. 3C) and fibrosis (Supporting Fig. 3D) were significantly decreased upon HO-1 induction. Quantitative evaluation and comparison of histological staining revealed that HO-1 induction, at late time points, improved inflammation and fibrosis to better scores than observed at the age of 12 weeks. This observation was statistically significant for portal inflammation (Fig. 5A) and lobular fibrosis (Fig. 5B).
HO-1 Induction Impedes Proliferation and Dysplasia in Mdr2ko Mice.
HCC frequency is increased in patients after years of chronic inflammation.29 Mdr2ko mice, which suffer from chronic hepatic inflammation, have been shown to develop HCC from the age of 12-15 month onward.12 To analyze effects of HO-1 induction on early signs of progression to HCC, we first investigated expression levels of growth factors and proliferation markers. TGF-β2 was significantly down-regulated in livers of Mdr2ko mice upon HO-1 induction during early (Fig. 6A) and established fibrosis (Fig. 6B). The proliferation marker, PCNA, and the cell-cycle regulators, cyclinD1 and E1, were down-regulated in 12-week-old CoPP-treated Mdr2ko mice, compared to solvent-treated mice (Fig. 6A). Likewise, HO-1 induction during already-established fibrosis significantly reduced the expression of cyclinD1 and PCNA (Fig. 6B). These results are supported by the finding that in liver slices of 12-week-old mice, significantly more PCNA-positive stained hepatocyte nuclei were found after solvent treatment (6,095 ± 203.88 PCNA nuclei/mm2), compared to CoPP-treatment (4,268.33 ± 175.94 PCNA nuclei/mm2; P < 0.05) (Supporting Fig. 4). These results indicate that the combination of anti-inflammatory and -fibrotic HO-1 effects might reduce the risk of Mdr2ko mice to progress to tumor formation. This hypothesis is supported by histological staining, revealing significantly less signs of dysplasia (e.g., irregular hepatic plates, hepatocellular enlargement, nuclear polymorphisms, imbalanced nucleus/cytoplasm ratio, and related to large-cell dysplasia) in CoPP-treated mice (Fig. 6C). For quantification, 12-week-old mice were solvent treated (0.41 ± 0.507), compared to CoPP-treated mice (0.07 ± 0.067; P < 0.05); also, 19-week-old mice were solvent treated (0.64 ± 0.497), compared to CoPP-treated mice (0.00 ± 0.000; P < 0.05).
Chronic inflammation, either caused by viral infections, alcoholic, or nonalcoholic steatohepatitis, parasites, or autoimmune diseases, frequently leads to persistent wound healing and fibrogenesis.30, 31 In animal models of acute hepatitis, HO-1 has been shown to protect from inflammatory liver damage via its products, CO and biliverdin.7, 8 We have investigated the anti-inflammatory effects of HO-1 in the Mdr2ko mouse model of chronic hepatitis, primary sclerosing cholangitis (PSC), and progression to HCC. In general, protection was characterized by reduced hepatic leukocyte infiltrations. This might have been the result of the fact that HO-1 induction interfered with the expression of OPN in the liver, a member of the small integrin-binding ligand N-linked glycoprotein family of proteins, which is an early marker of T-cell activation and is crucially involved in the recruitment of monocytes/macrophages, neutrophils, and natural killer T cells, thereby promoting inflammation.26
TNF is a known mediator of inflammatory processes.32 TNF signaling has been described for intestinal cells from patients with inflammatory bowel disease,33 as well as for hepatocytes, Kupffer cells, and infiltrating mononuclear cells from patients with chronic viral hepatitis.34 TNF is a crucial mediator of acute experimental hepatitis and has been shown to be down-regulated by HO-1 induction or overexpression.7 In our model of chronic hepatic inflammation, we observed increased TNF expression, but did not observe changes in TNF expression by HO-1 induction in whole liver tissue. On the other hand, TNF signaling seemed to be down-regulated as a result of reduced expression and higher shedding rates of its receptors, because soluble TNFRs have been shown to efficiently block TNF bioactivity.35 This is in line with recent findings showing down-regulation of TNFR1-dependent oxidative stress in cultured human tracheal smooth muscle cells upon HO-1 overexpression.30 In consequence, downstream signaling of TNF was affected by HO-1 induction, because we found significantly reduced levels of activated p38, Erk42/44, and activated Erk42 in CoPP-treated mice—all involved in TNF-dependent gene transcription.
Recent studies in a PSC and fibrosis model in OPN- or TNFR1-knockout mice showed that both factors are crucial for the establishment of 3,5-diethoxycarbonyl-1,4-dihydrocollidine–induced toxic fibrosis.36 Both OPN and TNFR1 were found to be down-regulated in our mouse model. We could show that HO-1 induction prevents the progression of fibrosis, indicated by, for example, lower levels of hydroxyproline, as well as decreased expression levels of collagens I and III, which both are highly present in fibrotic livers.37 Likewise, expressions of collagens I and III were found to be significantly reduced in CoPP-treated mice, even when fibrosis was already established. Antifibrotic activity of HO-1 could either be a consequence of reduced inflammation or mediated by a specific antifibrotic mechanism. Our findings are in line with reports showing that specific overexpression of HO-1 in HSCs attenuates CCl4-induced liver fibrosis in rats.38 We also found evidence that HO-1 induction might even resolve established fibrosis. The ratio between MMP and TIMP expression seems to be crucial for the switch from fibrosis progression to fibrolysis.39 We found that HO-1 induction directly interfered with HSC activation, whereas it induced the expression of MMP-9 and -13 more dramatically than the expression of TIMP1. The ratio of MMP-13/TIMP-1 has been linked to fibrolysis,40 and, in fact, we found fibrolytic activity caused by HO-1 induction. These findings suggest that HO-1 bears anti-inflammatory, antifibrotic, and fibrolytic capacity.
Fibrosis progression frequently results in the development of HCC.41 This process is accompanied by increased proliferation and growth-factor activity. TGF-β1, which is one of the most potent profibrogenic cytokines,42 has been linked to wound healing and carcinoma progression.43 Beside proliferation, TGF-β has been shown to be involved in differentiation of fibroblasts into myofibroblasts as well as in ECM production and deposition.43 TGF-β2 has been shown to be specifically expressed in epithelial cells of proliferating bile ducts in fibrotic livers of rats and humans44 and could, therefore, support PSC formation. We could show that TGF-β2 was significantly down-regulated in Mdr2ko mice upon HO-1 induction, independently of the time point of treatment onset, indicating that HO-1 induction would reduce bile duct proliferation and liver inflammation. Proliferation is further characterized by increased expression levels of cell-cycle regulators, such as cyclinD1 and E1, as well as PCNA.45, 46 Those markers were found to be down-regulated upon HO-1 induction, further indicating inhibition of proliferation. These results are in line with our observation that HO-1 induction reduced early signs of dysplasia and indicate that HO-1 induction at early time points (e.g., during inflammation or fibrosis) might have consequences on subsequent progression to HCC. Preliminary results even indicate that early HO-1 induction might interfere with progression to HCC.
HO-1 has been shown to be overexpressed in alcohol-induced HCC in patients.47 Moreover, HO-1 down-regulation via short interfering RNA significantly decreased tumor growth, whereas it increased cellular damage and apoptosis.47 Therefore, in the liver, HO-1 overexpression seems to exert beneficial or detrimental effects, depending on pathological conditions (e.g., inflammation or solid tumor formation). However, tumor-promoting effects by early HO-1 induction are unlikely, because induced HO-1 protein is degraded within days after treatment, whereas anti-inflammatory effects of HO-1 induction seem to last for at least 8 weeks longer than treatment with CoPP. Therefore, follow-up experiments have to determine the consequences of early HO-1 induction on progression to HCC caused by chronic inflammation.
The expert technical assistance by Elena Tassika, Christine Loscher, and Nicola Peters is gratefully acknowledged.