Stephan Hailfinger and Maike Jaworski contributed equally to this work.
Liver Biology and Pathobiology
Zonal gene expression in murine liver: Lessons from tumors†
Article first published online: 22 FEB 2006
Copyright © 2006 American Association for the Study of Liver Diseases
Volume 43, Issue 3, pages 407–414, March 2006
How to Cite
Hailfinger, S., Jaworski, M., Braeuning, A., Buchmann, A. and Schwarz, M. (2006), Zonal gene expression in murine liver: Lessons from tumors. Hepatology, 43: 407–414. doi: 10.1002/hep.21082
Potential conflict of interest: Nothing to report.
- Issue published online: 22 FEB 2006
- Article first published online: 22 FEB 2006
- Manuscript Accepted: 22 DEC 2005
- Manuscript Received: 25 OCT 2005
- Deutsche Forschungsgemeinschaft. Grant Number: SCHW 329/3-2
- Deutsche Krebshilfe. Grant Number: 106356
Gene expression in hepatocytes within the liver lobule is differentially regulated along the portal to central axis; however, the mechanisms governing the processes of zonation within the lobule are unknown. A model for zonal heterogeneity in normal liver is proposed, based on observations of differential expression of genes in liver tumors from mice that harbor activating mutations in either Catnb (which codes for β-catenin) or Ha-ras. According to the model, the regulatory control consists of two opposing signals, one delivered by endothelial cells of the central veins activating a β-catenin–dependent pathway (retrograde signal), the other by blood-borne molecules activating Ras-dependent downstream cascades (anterograde signal). In conclusion, gradients of opposing signaling molecules along the portocentral axis determine the pattern of enzymes and other proteins expressed in hepatocytes of the periportal and pericentral domains of the liver lobule. (HEPATOLOGY 2006;43:407–414.)
Even though the architecture of the liver tissue appears rather uniform, a closer view shows the existence of repetitive small structural and functional units called lobules that compose the liver parenchyma. Hepatocytes within the liver lobule differ in their enzyme content and subcellular structure according to their location relative to the afferent and efferent blood vessels, the terminal branches of the portal and the hepatic (central) veins, respectively (for reviews see1, 2). Based on the location of the blood vessels and the direction of the blood flow, the individual liver lobule can therefore be subdivided into an upstream “periportal” and a downstream “perivenous” (pericentral) region. Hepatocytes located in either of the two regions have different, often complementary functions, as indicated by differences in the content and activity of key enzymes of the intermediary and xenobiotic metabolism.
Two types of zonal patterns of gene expression have been described: one group of genes is stably expressed (non-inducible) within only one or a few layers of hepatocytes of the liver lobule, either pericentral or periportal, whereas a second group of genes displays a more dynamic (inducible) expression that may gradually diminish along the axis of the lobule.3, 4 One of the best-known examples for an enzyme encoded by a member of the first group of genes is glutamine synthetase (GS), which plays a key role in ammonia metabolism. GS is stably expressed at very high levels within only one to two layers of hepatocytes surrounding the central veins, and the number of hepatocytes expressing the enzyme is hardly affected by external stimuli under physiological conditions.5 Other enzymes, exemplified by key enzymes of carbohydrate metabolism, are also zonally expressed, but in a more gradual pattern. In addition, these latter enzymes undergo dynamic changes in expression as an adaptive response to changes in the hormonal or nutritional status. Xenobiotic metabolizing enzymes, including various cytochromes P450 (CYPs), represent an intermediate behavior: under normal conditions the expression of most CYPs is restricted to a small layer of hepatocytes located around the central veins; however, the expression extends further and further toward the periportal areas when animals are challenged with increasing doses of enzyme inducers such as phenobarbital, polychlorinated biphenyls, or 2,3,7,8-tetrachlorodibenzo-p-dioxin.6–8
Several hypotheses have been developed to explain the mechanisms underlying metabolic zonation in liver (for summary see4). Based on the observation that GS and ornithine aminotransferase, another enzyme of ammonia metabolism in liver, are exclusively expressed in hepatocytes contacting central (but not portal) veins, it was suggested that direct interaction of hepatocytes with central veins may trigger position-specific expression of the two genes.9 An alternative hypothesis suggests the existence of a portocentral gradient in the concentration of blood-borne signaling molecules, including oxygen, which results from the unidirectional blood flow within the liver lobule. As a consequence, interactions of these molecules with hepatocytes gradually decrease along the portal to central axis.4 The nature of the signaling molecules postulated to mediate zone-specific transcriptional responses, however, is unknown.
Previous work in our laboratory has been primarily focused on mechanisms underlying chemical hepatocarcinogenesis in rodents. In these studies, we demonstrated that experimentally induced liver tumors in mice harbor, depending on the treatment regimen, activating mutations in either Catnb or Ha-ras. In mice treated with a single injection of the liver carcinogen N-nitrosodiethylamine (DEN) at the age of 6 weeks, approximately 30% of liver tumors possess activating mutations in the Ha-ras oncogene. By contrast, if the carcinogen regimen is followed by continuous administration of the model tumor promoter phenobarbital (PB) through the diet, approximately 80% of tumors contain activating mutations in Catnb, which codes for the oncoprotein β-catenin.10 We recently reported that Catnb mutated tumors strongly express GS11 and several CYP isoforms,12 which, by contrast, are almost entirely absent in Catnb wild-type, presumably Ha-ras–mutated tumors. Subsequent microarray analysis of global gene expression demonstrated strong differences between tumors of the two genotypes, and several functional clusters were identified that involve changes in amino acid utilization and ammonia disposition in Catnb-mutated tumors, as opposed to alterations in lipid and cholesterol metabolism in Ha-ras–mutated tumors.13 Our results suggested that tumor cells in liver undergo highly coordinated alterations in a certain subset of genes orchestrated by signals from the activated Ha-Ras and β-catenin oncoproteins, respectively. In the course of this analysis, we noticed that the patterns of gene expression observed in tumors of the two genotypes resembled those seen in periportally and pericentrally located hepatocytes in normal liver tissue, suggesting that the same signaling pathways may be operative in normal liver cells. In this report we summarize our findings and develop a hypothesis to explain zonal heterogeneity of gene expression in liver.
Materials and Methods
Mouse liver samples were available to us from previous experiments conducted in our laboratory, where liver tumors had been induced by a single injection of DEN without further treatment14 or by combining DEN treatment with a subsequent continuous administration of the tumor promoter PB in the diet.10, 15 Frozen aliquots of liver tissue and isolated liver tumors were used for immunohistochemistry and biochemical examination.
Immunohistochemical stainings were performed by standard protocols as recently described.12 In brief, frozen liver sections were incubated with the respective primary antibodies (see below), and antibody binding was visualized by use of appropriate anti-IgG secondary antibodies conjugated with horseradish peroxidase (Sigma-Aldrich, Taufkirchen, Germany) and 3-amino-9-ethylcarbazole/H2O2 as substrates. The following antibodies and dilutions were used: mouse anti-GS monoclonal antibody (1:2,000; BD Biosciences Europe, Heidelberg, Germany), mouse anti-E-cadherin monoclonal antibody (1:400, Transduction Laboratories, Lexington, KY), rabbit antisera against CYP2C, GST-μ (1:300 and 1:500, both gifts of Dr. R. Wolf, Biomedical Research Centre, University of Dundee, Dundee, UK), and CYP2E1 (1:500, Stressgene, Victoria, Canada).
Frozen tissue aliquots were homogenized in lysis buffer [50 mmol/L Hepes, 150 mmol/L NaCl, and 1% Triton X-100, plus protease inhibitor cocktail (Complete Mini, Roche, Mannheim, Germany)] and protein concentrations were estimated using the Bradford assay. Western blots were performed as recently described,10 using antibodies against GS (1:5,000, BD Biosciences), E-cadherin (1:1,000, Transduction Laboratories), and GAPDH (1:500, Chemicon, Temecula, CA). Antibody binding was visualized using appropriate alkaline-phosphatase–conjugated secondary antibodies and CDP-Star as substrate (Tropix, PE Applied Biosystems, Weiterstadt, Germany).
Tumor tissue samples were taken from GS-stained serial liver sections by the use of punching cannuli16 and used for polymerase chain reaction (PCR) amplification of a 248-bp fragment containing exon 3 of the Catnb gene (forward: 5′-ACTCTGTTTTTACAGCTGACC-3′; reverse: 5′-TTTACCAGCTACTTGCTCTTG-3′) and a 166-bp fragment containing codon 61 of the Ha-ras gene (forward: 5′-GAGACATGTCTACTGGACATCTT-3′; reverse: 5′-GCTAGCCATAGGTGGCTCACCTG-3′). PCR products were analyzed by dideoxynucleotide sequencing (Seqlab Laboratories, Göttingen, Germany) of both strands using the amplification primers in the case of Catnb and one strand for Ha-ras using 5′-TGTTGTTGATGGCAAATACACAGAGG-3′. Each detected mutation was confirmed by independent PCR/sequencing reactions with at least 2 tissue samples taken from the same tumors in serial sections.
Isolation and Cultivation of Hepatocytes.
Hepatocytes were isolated from mouse liver by standard collagenase perfusion and seeded on collagen-coated dishes at a density of 70,000 cells/cm2 in Dulbecco's minimum essential medium/F-12 medium containing 10% fetal bovine serum and antibiotics. After 3 hours, medium was changed to 1% serum, and cells were incubated with 5 μmol/L glycogen synthase kinase inhibitor SB-216763 (Sigma) or solvent (0.05% dimethyl sulfoxide) for 96 hours. In a second experiment, cells were grown for 72 hours in medium conditioned by a 3T3 fibroblast cell line producing active Wnt3A (gift of Dr. R. Kemler, Freiburg, Germany). Medium conditioned by untransfected 3T3 cells was used as a control. Medium was changed daily. The experiments were performed in triplicate.
Periportal and perivenous subpopulations of hepatocytes were isolated by perfusion with digitonin/collagenase17 with minor modifications. Viability of the isolated hepatocytes was approximately 80% to 90% as determined by trypan blue staining. The experiment was performed in duplicate.
Quantitative Determination of mRNAs by Reverse Transcription Polymerase Chain Reaction.
Total RNA was isolated from hepatocyte preparations by Trizol reagent (Invitrogen, Karlsruhe, Germany). Five hundred nanograms RNA were reverse transcribed into cDNA by AMV-RT (Promega, Mannheim, Germany) using standard methods. Expression analysis was performed using the LightCycler real-time PCR system (Roche, Mannheim, Germany). Expression of β-actin was used for normalization. The primer pairs used were: E-cadherin, forward: 5′-TCTACCAAAGTGACGCTGAA-3′, reverse: 5′-GCTGATGGGAGGGATGAC-3′. H19, forward: 5′-CTGCTGCTCTCTGGATCCTC-3′, reverse: 5 ′-GTGGGTGGGTGCTATGAGTC-3′. Rex3, forward: 5′-AGGAGGAAGAGCGGAGCA-3′, reverse: 5 ′-AAGCTGGTAACAGGGAGAGATC-3′. CAR, forward: 5′-GCTGCCTAAGGGAAACAGGA-3 ′, reverse: 5′-AGCAAACGGACAGATGGGAC-3′. CYP1a1, forward: 5′-TGTCCTCCGTTACCTGCCTA-3′, reverse: 5′-GTGTCAAACCCAGCTCCAAA-3′. Rhbg, forward 5′-TACAACCACGAAACCGACG-3′, reverse: 5′-CAAACTCTCCACGCCAACA-3′. Axin2, forward: 5′-CGACGCACTGACCGACGATT-3′, reverse: 5′-TCCAGACTATGGCGGCTTTCC-3′. Gpr49, forward: 5′-AATCGCGGTAGTGGACATTC-3′, reverse: 5′-GATTCGGAAGCAAAAATGGA-3′. GS, forward: 5′-GCGAAGACTTTGGGGTGATA-3′, reverse: 5′- GTGCCTCTTGCTCAGTTTGTC-3′. β-Actin, forward: 5′-TCTGGCACCACACCTTCTACA-3′, reverse: 5′-GGGGTGTGTTGAAGGTCTCAAAC-3′. AhR, forward: 5′-GTCAAATCCTTCTAAGCGACACA-3′, reverse: 5′-AACCAGCACAAAGCCATTCA-3′. CYP2b10, forward: 5′-TACTCCTATTCCATGTCTCCAAA-3′, reverse: 5′-TCCAGAAGTCTCTTTTCACATGT-3′. CYP2e1, forward: 5′-TCCCTAAGTATCCTCCGTGA-3′', reverse: 5′-GTAATCGAAGCGTTTGTTGA-3′. We performed 4 measurements per mRNA isolated from hepatocytes of 2 independent preparations.
Recent work from our laboratory showed that Ha-ras and Catnb mutated mouse liver tumors display contrasting phenotypes with respect to a large variety of proteins/genes.13 Among these is GS, which is a transcriptional target of β-catenin signaling.11, 18 Because GS can be easily demonstrated by immunohistochemistry, it is a very suitable marker for the identification of liver lesions harboring an activated (mutated) version of β-catenin.11, 18 GS-positive lesions predominate in livers of mice treated with a regimen including PB as tumor promoter, but also can infrequently be observed in mice treated with DEN alone.11, 12 Most liver lesions from mice that were treated with DEN alone were GS-negative; however, a few, comparatively small GS-positive lesions were also identified. Tissue samples from GS-positive and GS-negative lesions were isolated from the stained sections and used for analysis of Catnb and Ha-ras mutations. Additional serial sections were stained immunohistochemically for a variety of selected other proteins that were chosen as “markers” because of their known zonal differences in expression within the liver lobule. The results of this study are exemplified in Fig. 1. The GS-positive lesion in Fig. 1 contained a heterozygous deletion in the Catnb gene, whereas the GS-negative lesion harbored a heterozygous point mutation in codon 61 of Ha-ras. In the immunohistochemical stainings, Catnb and Ha-ras mutated liver lesions demonstrated opposing staining reactions. GS-positive, Catnb mutated lesions expressed low levels of E-cadherin but were clearly positive for cytochromes P450 2C (CYP2C) and 2E1 (CYP2E1), and also for glutathione-S-transferase μ (GST-μ). By contrast, GS, CYPs, and GST-μ were almost undetectable in Ha-ras mutated lesions, and E-cadherin expression was increased (Fig. 1). GS and E-cadherin expression was also investigated by Western analysis of Catnb or Ha-ras mutated liver tumors (Fig. 2). Catnb mutated tumors expressed GS but lacked expression of E-cadherin, whereas Ha-ras mutated tumors lacked GS but showed expression of E-cadherin. In normal liver these proteins demonstrated a marked zonal heterogeneity in staining intensity (see Fig. 1). GS and the xenobiotic-metabolizing enzymes showed higher or exclusive expression in hepatocytes of the pericentral (c) domain, whereas E-cadherin was predominantly concentrated in hepatocytes surrounding the portal (p) venules. Therefore, with respect to these markers, Ha-ras mutated hepatoma cells resemble periportal hepatocytes, whereas Catnb mutated hepatoma cells resemble pericentral ones.
This dichotomy is further substantiated by observations from a recently completed study which investigated changes in global gene expression in Catnb and Ha-ras mutated mouse liver tumors.13 On examination of the battery of genes changed in expression in the two tumor types, we noticed that a variety of them code for proteins that are well known from the literature to show zonal differences in their expression within the liver lobule. In Table 1, we summarize data for a subset of genes with characteristic expression differences in Ha-ras and Catnb mutated tumors and compare these with literature data on zonal differences in the levels of the corresponding proteins in normal liver. Enzymes in Table 1 are grouped into classes according to their function within metabolic pathways. The results of this comparison are in good agreement with our hypothesis that Ha-ras and Catnb mutated hepatoma cells display phenotypes that resemble those of periportal and pericentral hepatocytes, respectively.
|Pathways/Enzymes/Proteins||pp||Ha-ras||pc||Catnb||Zonation in Liver||Changes in Tumors|
|CYP2E1||↓||↓||↑||↑||(2), this work||(12), this work|
|CYP2C||↓||↓||↑||↑||(6), this work||(12), this work|
|CYP (cytochrome) oxidoreductase||↓||↓||↑||↑||(6)||(13), (#)|
|GST mu (GSTc)||↓||↓||↑||↑||(6), this work||this work|
|Epoxide hydrolase 2, cytoplasmic||↓||↓||↑||-||(2)||(#)|
|Aryl hydrocarbon receptor (AhR)||↓||-||↑||↑||(29)||(13)|
|Constitutive androstane receptor (CAR)||↓||↓||↑||↑||(19), this work||(13) and unpublished observation|
|Glutamine synthetase||↓||↓||↑||↑||(2)||(13), this work|
|Carbonic anhydrase 2||↓||-||↑||↑||(2)||(13)|
|Carbonic anhydrase 3 and 5a||↓||↓||↑||-||(2)||(#)|
|Argininosuccinate synthetase 1||↑||-||↓||↓||(20)||(13)|
|Ornithine amino transferase||↓||↓||↑||↑||(21)||(18) (13)|
|Bile acid synthesis|
|Cholesterol 7-α-hydroxylase (CYP7A1)||↓||↓||↑||-||(2)||(13)|
|Cholesterol 26-hydroxylase (CYP27A1)||↓||↓||↑||-||(2)||(13)|
|E-cadherin||↑||↑||↓||↓||(23), this work||(#), this work|
In our recent study,13 several genes were significantly up- or down-regulated in Ha-ras and Catnb mutated tumors, for which a connection to tumorigenesis has only very recently been established, and it was not known whether they would also exhibit zonal heterogeneity in expression in normal liver. Using a technique by which hepatocytes from the periportal and pericentral domains are enriched by a combined digitonin/collagenase perfusion of the liver,17 pericentral and periportal subpopulations of hepatocytes were isolated and total RNA was extracted from the 2 cell populations to quantify the expression of various “marker” mRNAs by LightCycler analysis. The quality of the separation procedure was controlled by determination of GS mRNA levels, which were approximately 30-fold higher in pericentral as compared with periportal hepatocytes, indicating considerable enrichment in the respective subfractions. Comparable results were obtained by Western analysis and immunohistochemical staining for GS (not shown). The results obtained for a variety of additional genes are shown in Fig. 3. That the zonal expression patterns of the various marker mRNAs perfectly matched their expression in tumors of the respective genotype is remarkable: Genes that are strongly upregulated in Ha-ras mutated tumors (exemplified by Rex3 and H19) showed higher levels of expression in periportal hepatocytes, whereas genes that are strongly overexpressed in Catnb-mutated tumors (exemplified by Gpr49 and axin2/conductin) demonstrated preferential pericentral expression.
Physiologically, β-catenin–dependent gene transcription can be stimulated through Wnt-mediated activation of cell surface receptors of the Frizzled family. We have therefore incubated primary mouse hepatocytes with conditioned medium from a fibroblast Wnt3a producer line. In response to Wnt3a, the hepatocytes expressed significantly higher mRNA levels of various CYPs, Axin2, and Gpr49 (Fig. 4). By contrast, GS mRNA was not significantly increased (not shown). Similar results were obtained when cells were incubated with the glycogen synthase kinase inhibitor SB-216763, which also mediates β-catenin activation (Fig. 4).
We developed a hypothesis to explain zonal heterogeneity in gene expression in the liver. Specifically, we postulate that two diametrically opposing signals, each gradual in nature, namely, a β-catenin–dependent signal, presumably delivered by endothelial cells of the central veins, and a second signal generated by blood-borne molecules activating a Ras-dependent pathway, determine periportal- and pericentral-specific hepatocyte differentiation. The hypothesis is primarily based on findings of gene expression patterns in mouse liver tumors harboring activating mutations in Catnb and Ha-ras, which correspond to those of pericentral and periportal hepatocytes, respectively.
Beta-catenin, encoded by Catnb, is part of the canonical signal transduction pathway initiated by binding of Wnt/Wingless molecules to receptors of the Frizzled family of transmembrane receptors.25, 26 When complexed to T-cell factor (TCF)/lymphocyte enhancer factor transcription factors, β-catenin activates transcription of a battery of target genes while it inhibits expression of others (see http://www.stanford.edu/∼rnusse/wntwindow.html). We have now observed that several proteins that are over- or under-represented in mouse liver tumors in comparison with the surrounding normal liver tissue are also subject to zonal regulation within the normal liver tissue (Table 1). This is exemplified by immunohistochemical stainings for GS and several CYP isoforms that are expressed in Catnb-mutated hepatomas and pericentral hepatocytes but are largely absent in Ha-ras–mutated hepatomas and hepatocytes within the periportal domain of normal liver. By contrast, E-cadherin, known to be negatively regulated by β-catenin at the level of gene transcription,26 is down-regulated in Catnb–mutated hepatomas and overexpressed in Ha-ras–mutated hepatomas, perfectly corresponding to the low and high expression levels of the protein in pericentrally and periportally located hepatocytes in the normal liver tissue (Fig. 2). Finally, mRNA levels of selected genes, namely, Gpr49, Axin2/Conductin, Rex3, and H19, which had not been investigated for zonal expression before, were analyzed in enriched populations of periportal and pericentral hepatocytes, and the results obtained with respect to their preferential expression strongly support our hypothesis.
Neither the nature nor the source of the signaling molecule(s) that determine(s) gene expression in pericentral hepatocytes is known; however, speculating that endothelial cells of the central veins produce Wnt molecules that activate β-catenin signaling in neighboring hepatocytes is tempting. Evidence in favor of this idea comes from co-culturing experiments of hepatocytes with cells of the endothelial-like line RL-ET-14 conducted by Gebhardt's group that demonstrated that GS is inducible in hepatocytes surrounding endothelial cells in the co-culture model.27 Notably, induction of GS could be inhibited in a similar co-culture system with RL-ET-14 cells and fetal hepatocytes if β-catenin expression were silenced by small-interfering RNA, clearly demonstrating that functional β-catenin signaling is required for GS induction.28 The most direct evidence for an important role of β-catenin in the regulation of GS in normal liver comes from experiments with heterozygeous GS+/LacZmice carrying the LacZ gene in the GS locus.18 In this system, a positive β-galactosidase staining reaction indicates activation of the GS promoter/enhancer and control mice show, as expected, a restricted β-galactosidase staining in pericentral hepatocytes. Injection of mice with an adenoviral construct that mediates expression of an activated form of β-catenin throughout the liver, however, resulted in staining of most hepatocytes within the liver lobule, demonstrating that GS expression in liver is controlled by a β-catenin–dependent signal.18
CYPs involved in drug metabolism are also almost exclusively expressed by hepatocytes within the pericentral compartment, but in a less restricted manner than GS. The expression of many CYP isoforms is under the control of nuclear receptors such as the constitutive androstane receptor (CAR) in the case of the phenobarbital inducible forms or the aryl hydrocarbon receptor (AhR) in the case of the 3-methylcholanthrene–inducible forms. We recently showed that transient expression of activated β-catenin in mouse hepatoma cells in culture stimulates CAR- and AhR-responsive luciferase reporters, suggesting that a β-catenin–dependent signal is involved in the transcriptional activation of the respective CYP forms.12 Interestingly, the concentration of both receptors is also changed in mouse liver tumors (see Table 1). AhR mRNA and protein levels are higher in pericentral than in periportal hepatocytes, corresponding to the differences in concentration of CYP1A subfamily enzymes in hepatocytes of the two domains.29 A similar zonal difference in mRNA level of CAR has been observed in the current study (see Fig. 3).
Hepatocytes from GS-negative tumors, which are often Ha-ras or B-raf mutated,30 never demonstrate a positive GS staining reaction, even if they are located directly adjacent to blood vessels (exemplified in Fig. 1; see also Loeppen et al.11). This is in contrast to the situation in normal liver and explains why GS mRNA and protein were decreased in homogenates from Ha-ras mouse liver tumors relative to normal liver tissue13 (Fig. 2). Based on our hypothesis, two alternative explanations may be proposed for this behavior: either endothelial cells of hepatic venules within tumors do not produce signals acting through β-catenin like their counterparts in the normal liver tissue, or hepatocytes with constitutive activation of the MAPK signal transduction pathway (because of a Ha-ras or B-raf mutation, or an alternative mutation in an unknown gene coding for a signaling molecule within the pathway) do no longer respond to the β-catenin activating signal. The latter alternative implies that Ras signals are dominant and capable of silencing β-catenin/TCF-mediated differentiation programs in liver. In fact, experimental evidence exists in favor of the idea of a silencing effect mediated by Ras proteins from published data: (i) expression of an activated version of Ha-ras in mammary carcinoma cells and in keratinocytes has been demonstrated to down-regulate AhR function and expression of CYP1A131; (ii) most GS-negative tumors from acutely PB-induced mice do not respond with expression of CYPs of family 2 as do their GS-positive counterparts12; (iii) PB-induction of responsive CYPs in isolated hepatocytes requires serum-free medium, which is likely attributable to the presence of growth factors such as epidermal growth factor in serum, which are known to activate Ras-dependent signaling and also suppress expression of CYP1A132 as well as CYP2B1 and CYP3A1.33 Very recently, a silencer element within intron 1 of GS was shown to mediate negative regulation of gene expression in periportally located hepatocytes,34 which could be subject to regulation by a Ras-dependent signal. This hypothesis contrasts results obtained in other systems: for example, Ras and β-catenin were found to cooperate in activation of the cyclin D1 promoter,35 and transcription of an artificial β-catenin/TCF-dependent reporter was stimulated by insulin and insulin-like growth factor-1, requiring activation of Ras.36 Potentially, inhibitory or co-stimulatory effects can be produced by simultaneous activation of the 2 signaling pathways, depending on the cellular context and the promoter under investigation.
Change in the source of the afferent hepatic blood induced by microsurgical portocaval transposition does not significantly alter the patterns of gene expression within the liver lobule.37 Because this manipulation does not change the direction of the transhepatic blood flow, hepatocytes in the periportal (upstream) domain still sense the highest concentrations of a hypothetical blood-borne signaling molecule that determines zonation of gene expression. We now suggest that this hypothetical molecule may act through a Ras-dependent signal transduction pathway. In addition, the results with GS+/LacZmice,18 the co-culture experiments with hepatocytes and cells of line RL-ET-14,27, 28 and our current data strongly suggest that a second, positional signal, potentially delivered from endothelial cells, activates β-catenin–dependent signaling in neighboring hepatocytes, and thereby determines gene expression profiles that are characteristic for hepatocytes of the pericentral domain. Our results obtained with conditioned medium from Wnt3a producer cells also favor this general hypothesis, and further strong evidence comes from a study published in this issue,39 which demonstrates that several enzymes involved in ammonia metabolism and various CYPs are not expressed in livers of mice with a liver-specific conditional knockout of β-catenin. We conclude that Ha-ras and Catnb mutated hepatoma cells “simply” recapitulate gene expression programs, which are characteristic for hepatocytes in the periportal and pericentral domains of the normal liver tissue.
We greatly acknowledge the excellent technical assistance by Elke Zabinsky and Sylvia Vetter. We also thank Dr. Rolf Kemler for gift of Wnt3a cells, Dr. Roland Wolf for gift of antibody and Drs. Howard Glauert and Larry Robertson for helpful discussions.
- 6Characterization, localization and regulation of a novel phenobarbital-inducible form of cytochrome P450, compared with three further P450-isoenzymes, NADPH P450-reductase, glutathione transferases and microsomal epoxide hydrolase. Carcinogenesis 1984; 5: 993–1001., , , , , , et al.
- 17Isolation of periportal and pericentral hepatocytes. In: PhillipsIR, ShephardEA, eds. Methods in Molecular Biology: Cytochrome P450 protocols. Volume 108. Totowa, NJ: Humana Press Inc, 1998: 319–328.: