Drug Metabolism and Disposition Fast Forward
First published on January 12, 2007; DOI: 10.1124/dmd.106.013656
0090-9556/07/3504-503-507$20.00
DMD 35:503-507, 2007
SHORT COMMUNICATION
Zonal Gene Expression in Mouse Liver Resembles Expression Patterns of Ha-ras and ß-Catenin Mutated Hepatomas
Albert Braeuning,
Carina Ittrich,
Christoph Köhle,
Albrecht Buchmann, and
Michael Schwarz
Institute of Pharmacology and Toxicology, Department of Toxicology, University of Tuebingen, Tuebingen, Germany (A.Br., C.K., A.Bu., M.S.); and Central Unit of Biostatistics, German Cancer Research Center, Heidelberg, Germany (C.I.)
(Received November 3, 2006;
accepted December 28, 2006)
 |
Abstract
|
|---|
Hepatocytes of the periportal and perivenous zones of the liver lobule differ in their levels and activities of various enzymes and other proteins. We have recently suggested that ß-catenin- and Ras-dependent signaling pathways play an important role in the regulation of perivenous and periportal gene expression profiles. This hypothesis was primarily based on similarities in zonal differences in gene expression of hepatocytes from normal liver with gene expression patterns of liver tumors: several proteins and mRNAs preferentially expressed in periportal hepatocytes were often overexpressed in Ha-ras mutated mouse liver tumors, whereas perivenous markers were overexpressed in Ctnnb1 (encoding ß-catenin) mutated tumors. We have now extended this work by use of data from two previously conducted microarray analyses aimed to analyze 1) global gene expression patterns of Ha-ras and Ctnnb1 mutated mouse liver tumors and 2) transcriptome differences between periportal and perivenous mouse hepatocytes. By comparison of the datasets, 134 genes or expressed sequences were identified that were present in both datasets. Gene expression patterns in perivenous hepatocytes and Ctnnb1 mutated hepatoma cells were strongly correlated: 96.5% of the genes present in both datasets were regulated in the same direction. In analogy, expression of 74.1% of the genes deregulated in Ha-ras mutated tumors was correlated with the respective expression patterns in periportal hepatocytes. These findings favor the hypothesis that gene expression patterns in periportal and perivenous hepatocytes are regulated, at least in part, by Ras- and ß-catenin-dependent signaling pathways.
Based on the location of the blood vessels, the terminal branches of the portal and the hepatic (central) veins, and the direction of the blood flow, the individual liver lobule can 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 several metabolic key enzymes (for review see Jungermann and Katz, 1989
; Gebhardt, 1992). This so-called "zonation" of metabolism in the liver is also of particular impact on the metabolism of pharmaceuticals and xenobiotics, since most of the enzymes involved in the metabolism of such compounds show distinct differences in basal as well as inducible expression between perivenous and periportal hepatocytes, with the main detoxification enzymes of both phase I and phase II of xenobiotic metabolism being located preferentially in perivenous hepatocytes (for reviews on zonation of xenobiotic metabolism see Jungermann and Katz, 1989; Oinonen and Lindros, 1998). As a consequence, various exogenous compounds lead to a preferential damage of hepatocytes located in either the perivenous or periportal area of the liver lobule (for review see Lindros, 1997).
We have recently developed a hypothesis to explain the mechanisms that govern zonal differences in gene expression in liver (Hailfinger et al., 2006). According to our hypothesis, two opposing signals, each gradual in nature, dictate the gene expression patterns of hepatocytes within the different areas of the liver lobule. We postulated that a ß-catenin-activating signal, which is probably delivered by endothelial cells of the central veins, and a second opposing signal, presumably generated by blood-borne molecules activating a Ras-dependent signaling pathway, trigger perivenous- and periportal-specific mRNA expression profiles. This concept was deduced from the observation of striking similarities in the expression of selected "marker" genes in Ctnnb1 (encoding ß-catenin) and Ha-ras mutated hepatoma cells with their corresponding expression patterns in perivenous and periportal hepatocytes, respectively. We have now extended our previous study and compared the transcriptome of periportal and perivenous hepatocyte subpopulations with that of Ha-ras and Ctnnb1 mutated mouse liver tumors.
|
Materials and Methods
|
|---|
Datasets on global mRNA expression patterns in mouse liver tumors mutated in either Ha-ras or Ctnnb1 and datasets on global mRNA expression patterns in perivenous and periportal mouse hepatocytes were available to us from previous experiments conducted in our laboratory (Stahl et al., 2005a; Braeuning et al., 2006). In these studies, gene expression profiles had been analyzed by use of the Affymetrix GeneChip MOE-430A (Affymetrix, Santa Clara, CA) containing approximately 22,600 probe sets including more than 14,000 well characterized mouse genes. Since the criteria for statistical analysis of expression data varied in the two studies, we reevaluated the tumor datasets using as discriminators a log2 expression ratio 
1 (equivalent to 2-fold change) and a threshold of 0.1 for the false discovery rate-adjusted p values to allow a direct comparison with our data on zonal differences in gene expression in normal liver (for further details see Stahl et al., 2005a,b; Braeuning et al., 2006). The former cutoff was chosen because expression differences smaller than 2-fold are very difficult to detect by quantitative reverse transcription-polymerase chain reaction, which was used in the previous studies for verification of the microarray results. The latter was chosen to keep the expected proportion of false positives below 10%.
|
Results
|
|---|
We have recently determined by microarray analysis the mRNA expression profiles of mouse liver tumors harboring activating mutations in either Ha-ras or Ctnnb1 (Stahl et al., 2005a). In that study, liver tumors of the two genotypes were found to differ strikingly with respect to their gene expression patterns. More recently, we performed a microarray analysis of periportal and perivenous subpopulations of mouse hepatocytes enriched by combined digitonin/collagenase perfusion (Braeuning et al., 2006). In this latter study, 198 genes or expressed sequences (corresponding to 243 probe sets) were identified that demonstrated a 2-fold difference in expression between hepatocytes from the two different zones of the liver at a threshold of 0.1 for the false discovery rate-adjusted p values. Since more stringent criteria for statistical significance had been applied in our tumor microarray study, we now reanalyzed the tumor datasets using the same cutoffs as in our study on gene expression in perivenous and periportal hepatocyte subpopulations. Under these conditions, 777 (Ctnnb1) and 1063 (Ha-ras) probe sets were identified representing transcripts that differed in expression from normal liver (for illustration see Fig. 1). We then used the tumor datasets for a comparison with the gene expression profiles in perivenous and periportal hepatocytes.
Upon overlay of the datasets, we found that 158 of the 243 probe sets (corresponding to 134 of 198 genes or expressed sequences) identified as differentially regulated in the zonation experiment were also present in either one of the two tumor datasets (see Fig. 1). A total of 55 probe sets complied with the condition: zonation of expression of the respective transcripts in normal liver and deregulation in Ctnnb1 but not Ha-ras mutated tumors. The analogous number for alterations in Ha-ras but not Ctnnb1 mutated tumors was 64. Zonation in normal liver and deregulation in both tumor types was indicated by 39 probe sets. Therefore, 94 (55 + 39) and 103 (64 + 39) probe sets were identified for which their transcripts show zonal expression in normal liver and alterations in Ctnnb1 or Ha-ras mutated tumors, respectively.
The tumor versus normal liver expression ratios of the 94 transcripts that showed alterations in Ctnnb1 mutated tumors and zone-specific expression were then plotted against their log2 expression ratios in perivenous versus periportal hepatocytes (Fig. 2A). Each dot represents one probe set. Dots in the upper right gray area indicate probe sets that were 1) up-regulated in Ctnnb1 mutated tumors and 2) showed higher expression in perivenous than in periportal hepatocytes. In analogy, dots in the lower left area represent probe sets down-regulated in Ctnnb1 mutated tumor cells with lower expression in perivenous than in periportal hepatocytes. Since several genes were represented by more than one probe set on the microarray, the numbers downsize from 94 to 85 if genes and expressed sequences are regarded instead of probe sets (for a detailed list see Supplemental Data Table 1). Of these 85 genes, 82 (96.5%) were regulated in the same direction, and only 3 genes (representing the four probe sets in the white areas of the plot in Fig. 2A) showed alterations in opposite directions.
View larger version (19K):
[in this window]
[in a new window]
|
FIG. 2. Gene expression patterns of perivenous and periportal hepatocytes from normal liver correlate with those of Ctnnb1 and Ha-ras mutated liver tumors. Using the microarray datasets and criteria of analysis described in Fig. 1, differential gene expression in perivenous and periportal hepatocytes was correlated with expression changes in Ctnnb1 or Ha-ras mutated liver tumors. The plots show log2 expression ratios of probe sets representing transcripts with significant alterations in both screens. A, gene expression ratios in Ctnnb1 mutated versus normal liver (n.l.) plotted against expression ratios in perivenous versus periportal hepatocytes. B, gene expression ratios in Ha-ras mutated versus normal liver (n.l.) plotted against expression ratios in periportal versus perivenous hepatocytes. Dots in the gray areas indicate probe sets with unidirectional alterations.
|
|
View larger version (27K):
[in this window]
[in a new window]
|
FIG. 3. Concordance of gene expression in Ctnnb1 and Ha-ras mutated hepatoma cells with the expression of the respective genes in perivenous and periportal normal hepatocytes. For comparison, genes were categorized according to the function of their products, as indicated. Numbers in the bars indicate the respective number of genes represented by the bar.
|
|
An analogous comparison was performed with gene expression data from Ha-ras mutated tumors, but for the sake of clarity, the y-axis was inverted, now showing expression ratios of periportal versus perivenous hepatocytes (Fig. 2B). The 103 probe sets, which indicated deregulated genes and expressed sequences in Ha-ras mutated tumors that also demonstrated significant zonal expression in normal liver, correspond, again, to 85 different transcripts (for a detailed list see Supplemental Data Table 2). Of these, 63 (74.1%) showed unidirectional changes in their expression in periportal hepatocytes and Ha-ras mutated tumors. Taken together, these results indicate that the gene expression patterns in the perivenous hepatocytes are highly correlated with those in Ctnnb1 mutated hepatoma cells, whereas the expression patterns of periportal hepatocytes show some correlation with those of Ha-ras mutated hepatoma cells.
In the Supplemental Data Tables 1 and 2, genes are categorized according to the presumed function of their proteins. Since we were interested to know whether the above described concordance between gene expression in periportal and perivenous hepatocytes and Ha-ras and Ctnnb1 mutated tumors would differ between functional categories, we performed gene expression profile comparisons for selected subsets of categorized genes as shown in Fig. 3. Within these categories we found up to 100% concordance of gene expression between perivenous hepatocytes and Ctnnb1 mutated tumors and between periportal hepatocytes and Ha-ras mutated tumors, respectively. However, certain genes with preferential periportal expression, belonging to the categories xenobiotic metabolism and transport, were decreased in Ha-ras mutated tumors, thus breaking the rule (for a list of these genes, see Supplemental Data Table 2).
With 24 members (22 metabolic enzymes plus 2 nuclear receptors mediating the induction of xenobiotic-metabolizing enzymes, i.e., aryl hydrocarbon receptor and constitutive androstane receptor), the genes associated with xenobiotic metabolism constitute the largest group of functionally related genes that exhibit significant expression differences between hepatocytes from the two zones of the liver, with 19 of them showing a preferentially perivenous zonation and 5 being mainly expressed in periportal hepatocytes (Braeuning et al., 2006). As shown in Table 1, 9 of these 24 genes were deregulated in both Ctnnb1 and Ha-ras mutated tumors and 13 genes were present in one of the tumor datasets, whereas only 2 genes were not aberrantly expressed in either of the two tumor genotypes. Upon comparison of expression profiles, it becomes apparent that most perivenous genes of both phase I and phase II of xenobiotic metabolism are overexpressed in Ctnnb1 mutated tumor tissue (12 of 19 genes). On the other hand, in Ha-ras mutated hepatoma cells, the majority of the perivenous genes is down-regulated (11 of 19) with only one exception breaking the rule. Looking at the few genes with periportal localization, a general down-regulation in Ctnnb1 mutated tumors (three of five) is found. Ha-ras mutated tumors also express lower levels of mRNAs for periportal enzymes (three of five), again, with one gene being regulated in the opposite direction. Thus, most of the periportally localized genes of the xenobiotic metabolism fail to show a correlation between Ha-ras mutated tumors and periportal hepatocytes.
View this table:
[in this window]
[in a new window]
|
TABLE 1 Zonal-specific expression of xenobiotic-metabolizing enzymes in the liver and correlation to Ctnnb1 and Ha-ras mutated liver tumors
Significant (adjusted p value  0.1) up- or down-regulation 2-fold is indicated by arrows; arrows in parentheses indicate a difference in expression of <2-fold; dashes indicate lack of significant change. Perivenous zonation is highly correlated to overexpression in Ctnnb1 mutated tumors and down-regulation in Ha-ras mutated hepatomas.
|
|
|
Discussion
|
|---|
Recently, we have developed a hypothesis to explain zonal heterogeneity in gene expression in murine liver (Hailfinger et al., 2006). We postulated the existence of two diametrically opposing signals, each gradual in nature, one activating ß-catenin-dependent and the other activating Ras-dependent downstream targets, which together determine periportal- and perivenous-specific hepatocyte differentiation. This hypothesis was primarily based on the finding that expression patterns of selected marker proteins/mRNAs in mouse liver tumors with activating mutations in Ctnnb1 or Ha-ras were very similar to those of perivenous and periportal hepatocyte populations, respectively. The results from our present investigation on global gene expression profiles considerably extend our previous findings and give further support in favor of our recent hypothesis.
Our idea that zonal-specific gene expression in liver is triggered by external stimuli and is not intrinsically defined in periportal and perivenous hepatocytes is supported by cell culture experiments, where hepatocytes are removed from their physiological environment, such as the blood stream and neighboring endothelial cells. Under these conditions, striking differences in expression of zonated marker proteins such as GS or Cyp1a already begin to alleviate a few hours after perfusion (unpublished observation). Furthermore, addition of serum to hepatocyte cultures strongly attenuates expression and inducibility of perivenous markers (unpublished observation), suggesting that a yet unidentified, probably Ras-activating serum factor may play a role. Further support for our hypothesis comes from the finding that the expression of perivenous marker mRNAs is clearly inducible in periportal hepatocyte subpopulations in vivo and in vitro by activation of ß-catenin signaling (Cadoret et al., 2002; Benhamouche et al., 2006; our unpublished observations).
There is good evidence for the idea that the "perivenous phenotype" of hepatocytes is mediated via signaling through ß-catenin since, for example, expression of an activated form of ß-catenin in transgenic mice results in positive staining for the perivenous marker GS in most hepatocytes within the liver lobule (Cadoret et al., 2002), whereas GS and various cytochrome P450 isoforms are not expressed in perivenous hepatocytes from mice with a liver-specific conditional knockout of ß-catenin (Sekine et al., 2006). In addition, the tumor suppressor adenomatous polyposis coli, an important regulator of ß-catenin signaling, was established as the "zonation keeper" in liver (Benhamouche et al., 2006). Activation of ß-catenin in perivenous hepatocytes may be triggered by endothelial cells of the central veins, which may deliver Wnt molecules activating upstream receptors within the Wnt/ß-catenin pathway. Endothelial cells play a decisive role in promotion of hepatic specification during the early stages of liver organogenesis (Matsumoto et al., 2001). Hepatocytes cocultured with cells of the endothelial-like line RL-ET-14 express GS when neighboring the endothelial cells (Gebhardt et al., 1998), and GS expression could be inhibited in this system by silencing of ß-catenin expression by small-interfering RNA (Kruithof-de Julio et al., 2005).
ß-Catenin signaling also seems to be an important regulator of xenobiotic metabolism in liver, first, because expression of various cytochrome P450 isoforms is absent in perivenous hepatocytes from mice with a liver-specific knockout of ß-catenin (Sekine et al., 2006); second, because activation of ß-catenin signaling is able to induce the expression of several cytochrome P450 isoforms in cultured primary mouse hepatocytes (Hailfinger et al., 2006); and third, because mutational activation of Ctnnb1 in liver tumors is able to induce an expression profile of xenobiotic-metabolizing enzymes that is highly correlated to that of perivenous hepatocytes (Table 1; see also Loeppen et al., 2005). On the other hand, in Ha-ras mutated tumor cells we observed a general down-regulation of perivenous mRNAs encoding xenobiotic metabolism-related enzymes. This is in line with the observation that an activated version of Ha-ras down-regulates aryl hydrocarbon receptor function and Cyp1a expression in mammary carcinoma cells and keratinocytes (Reiners et al., 1997). Often, enzymes that were up-regulated in Ctnnb1 mutated tumors were down-regulated in Haras mutated hepatomas, suggesting an antagonism of both pathways in the regulation of gene expression. However, since some periportal mRNAs were also down-regulated in Ha-ras mutated liver tumors, additional signaling cascades different from the ß-catenin or Ha-ras pathway seem to be involved in the regulation of these genes. A candidate signal transducer mediating the periportal phenotype of xenobiotic-metabolizing enzymes may be growth hormone, since it has been demonstrated to determine the preferential periportal localization of Cyp2c7 in rat liver (Oinonen et al., 2000).
In summary, our present data clearly demonstrate that the patterns of gene expression in perivenous hepatocytes strongly resemble those of Ctnnb1 mutated hepatoma cells with activated ß-catenin, whereas the gene expression patterns of periportal hepatocytes resemble those of Ha-ras mutated hepatoma cells. There is convincing evidence from this study and previous ones by our group and others (Cadoret et al., 2002; Loeppen et al., 2002; Benhamouche et al., 2006; Sekine et al., 2006) that ß-catenin signaling plays a decisive role in mediating the perivenous phenotype of hepatocytes, but further studies are required to identify the nature and the signal transduction mechanisms of the opposing factor(s) that appears to dictate the periportal hepatocyte phenotype.
|
Footnotes
|
|---|
This study was supported by the Deutsche Krebshilfe (Grant 106356).
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.106.013656.
The online version of this article (available at http://dmd.aspetjournals.org) contains supplemental material. 
Address correspondence to: Dr. Michael Schwarz, Institute of Pharmacology and Toxicology, Department of Toxicology, University of Tuebingen, Wilhelmstr. 56, 72074 Tuebingen, Germany. E-mail: michael.schwarz{at}uni-tuebingen.de
|
References
|
|---|
Benhamouche S, Decaens T, Godard C, Chambrey R, Rickman DS, Moinard C, Vasseur-Cognet M, Kuo CJ, Kahn A, Perret C, et al. (2006) Apc tumor suppressor gene is the "zonation-keeper" of mouse liver. Dev Cell 10: 759–770.[CrossRef][Medline]
Braeuning A, Ittrich C, Köhle C, Hailfinger S, Bonin M, Buchmann A, and Schwarz M (2006) Differential gene expression in periportal and perivenous mouse hepatocytes. FEBS J 273: 5051–5061.[CrossRef][Medline]
Cadoret A, Ovejero C, Terris B, Souil E, Levy L, Lamers WH, Kitajewski J, Kahn A, and Perret C (2002) New targets of beta-catenin signaling in the liver are involved in the glutamine metabolism. Oncogene 21: 8293–8301.[CrossRef][Medline]
Gebhardt R (1992) Metabolic zonation of the liver: regulation and implications for liver function. Pharmacol Ther 53: 275–354.[CrossRef][Medline]
Gebhardt R, Schuler M, and Schorner D (1998) The spontaneous induction of glutamine synthetase in pig hepatocytes cocultured with RL-ET-14 cells is completely inhibited by trijodothyronine and okadaic acid. Biochem Biophys Res Commun 246: 895–898.[CrossRef][Medline]
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.[CrossRef][Medline]
Jungermann K and Katz N (1989) Functional specialization of different hepatocyte populations. Physiol Rev 69: 708–764.[Abstract/Free Full Text]
Kruithof-de Julio M, Labruyere WT, Ruijter JM, Vermeulen JL, Stanulovic V, Stallen JM, Baldysiak-Figiel A, Gebhardt R, Lamers WH, and Hakvoort TB (2005) The RL-ET-14 cell line mediates expression of glutamine synthetase through the upstream enhancer/promoter region. J Hepatol 43: 126–131.[CrossRef][Medline]
Lindros KO (1997) Zonation of cytochrome P450 expression, drug metabolism and toxicity in liver. Gen Pharmacol 28: 191–196.[Medline]
Loeppen S, Koehle C, Buchmann A, and Schwarz M (2005) A beta-catenin-dependent pathway regulates expression of cytochrome P450 isoforms in mouse liver tumors. Carcinogenesis 26: 239–248.[Abstract/Free Full Text]
Loeppen S, Schneider D, Gaunitz F, Gebhardt R, Kurek R, Buchmann A, and Schwarz M (2002) Overexpression of glutamine synthetase is associated with beta-catenin-mutations in mouse liver tumors during promotion of hepatocarcinogenesis by phenobarbital. Cancer Res 62: 5685–5688.[Abstract/Free Full Text]
Matsumoto K, Yoshitomi H, Rossant J, and Zaret KS (2001) Liver organogenesis promoted by endothelial cells prior to vascular function. Science (Wash DC) 294: 559–563.[Abstract/Free Full Text]
Oinonen T and Lindros KO (1998) Zonation of hepatic cytochrome P-450 expression and regulation. Biochem J 329 (Pt 1): 17–35.
Oinonen T, Ronis M, Wigell T, Tohmo K, Badger T, and Lindros KO (2000) Growth hormone-regulated periportal expression of CYP2C7 in rat liver. Biochem Pharmacol 59: 583–589.[CrossRef][Medline]
Reiners JJ Jr, Jones CL, Hong N, Clift RE, and Elferink C (1997) Downregulation of aryl hydrocarbon receptor function and cytochrome P450 1A1 induction by expression of Ha-ras oncogenes. Mol Carcinog 19: 91–100.[CrossRef][Medline]
Sekine S, Lan BYA, Bedolli M, Feng S, and Hebrok M (2006) Liver-specific loss of ß-catenin blocks glutamine synthesis pathway activity and cytochrome P450 expression in mice. Hepatology 43: 817–825.[CrossRef][Medline]
Stahl S, Ittrich C, Marx-Stoelting P, Kohle C, Altug-Teber O, Riess O, Bonin M, Jobst J, Kaiser S, Buchmann A, et al. (2005a) Genotype-phenotype relationships in hepatocellular tumors from mice and man. Hepatology 42: 353–361.[CrossRef][Medline]
Stahl S, Ittrich C, Marx-Stoelting P, Kohle C, Ott T, Buchmann A, and Schwarz M (2005b) Effect of the tumor promoter phenobarbital on the pattern of global gene expression in liver of connexin32-wild-type and connexin32-deficient mice. Int J Cancer 115: 861–869.[CrossRef][Medline]
![[Feedback]](/icons/banner/feedbackACT.gif)
![[For Subscribers]](/icons/banner/subscriberACT.gif)
![[Archive]](/icons/banner/archiveACT.gif)
![[Search]](/icons/banner/searchACT.gif)
![[Contents]](/icons/banner/tocACT.gif)
Top
Abstract
Materials and Methods
