C-Myc and its target FoxM1 are critical downstream effectors of constitutive androstane receptor (CAR) mediated direct liver hyperplasia

Authors

  • William E. Blanco-Bose,

    1. Genetics & Stem Cell Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), ISREC—Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland
    Search for more papers by this author
  • Mark J. Murphy,

    1. Genetics & Stem Cell Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), ISREC—Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland
    Search for more papers by this author
  • Armin Ehninger,

    1. Genetics & Stem Cell Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), ISREC—Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland
    Search for more papers by this author
  • Sandra Offner,

    1. Genetics & Stem Cell Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), ISREC—Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland
    Search for more papers by this author
  • Christelle Dubey,

    1. Genetics & Stem Cell Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), ISREC—Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland
    Search for more papers by this author
  • Wendong Huang,

    1. Department of Gene Regulation and Drug Discovery, Beckman Research Institute of City of Hope, Duarte, CA
    Search for more papers by this author
  • David D. Moore,

    1. Department of Molecular and Cellular Biology, Baylor College of Medicine, Houston, TX
    Search for more papers by this author
  • Andreas Trumpp

    Corresponding author
    1. Genetics & Stem Cell Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), ISREC—Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland
    2. Department of Cell Biology, German Cancer Research Center (DKFZ), Heidelberg, Germany
    • German Cancer Research Center (DKFZ), D-69120 Heidelberg, Germany
    Search for more papers by this author
    • A.T. is member of the EMBO Young Investigator Program.

    • fax: (49)-6221-422840.


  • Potential conflict of interest: Nothing to report.

Abstract

In the adult liver, 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP), an agonist of the constitutive androstane receptor (CAR, NR1I3), produces rapid hepatomegaly in the absence of injury. In this study, we identify c-Myc as a gene induced by CAR and demonstrate that TCPOBOP-induced proliferation of hepatocytes depends on c-Myc function. Moreover, the TCPOBOP-induced cell cycle program (Cdc2, cyclins, MCM proteins, Cdc20, and genes implicated in the spindle assembly checkpoint) is severely impaired in c-Myc mutant livers. Strikingly, many of these genes overlap with a program controlled by the forkhead transcription factor FoxM1, known to control progression through S-phase and mitosis. Indeed, FoxM1 is also induced by TCPOBOP. Moreover, we show that c-Myc binds to the FoxM1 promoter in a TCPOBOP-dependent manner, suggesting a CAR → c-Myc → FoxM1 pathway downstream of TCPOBOP. Conclusion: Collectively, this study identifies c-Myc and FoxM1 mediated proliferative programs as key mediators of TCPOBOP-CAR induced direct liver hyperplasia. (HEPATOLOGY 2008.)

The liver consists predominantly of differentiated hepatocytes that perform important metabolic functions. Among the functions of the liver are the detoxification of harmful xenobiotics by cytochrome P450s and other enzymes.1 This transformation process also can lead to an increase in the metabolism and elimination of therapeutic agents. In addition, some xenobiotics either themselves undergo transformation or cause other drugs to be transformed into toxins and carcinogens.2

The liver responds to the presence of xenobiotics by inducing members of the nuclear hormone receptor superfamily, particularly the pregnane X receptor and the constitutive androstane receptor (CAR).3, 4 An ideal candidate for studying xenobiotic metabolism is the halogenated hydrocarbon 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP), which is a strong agonist of CAR.4 TCPOBOP is both a nongenotoxic carcinogen on its own and a potent tumor promoter when combined with genotoxic agents.5 It potently induces genes associated with xenobiotic detoxification and hepatocyte proliferation in a CAR-dependent manner.2, 6–8 The effectors downstream of CAR responsible for mediating its proliferative and carcinogenic effects remain largely undefined,9 although mice deficient in Mdm2, a direct target of CAR, have a compromised proliferative response.8 It also has been hypothesized that the transcription factor FoxM1, which has been shown to promote S and M phase entry and the proper execution of mitosis, may play an important role in this process.7, 10, 11

Because the TCPOBOP-induced hyperplastic response involves both growth and proliferation of hepatocytes, a potentially important protein in this process may be c-Myc, which has been implicated in various aspects of liver proliferation such as that observed in liver regeneration12 (Blanco-Bose W, Murphy M, Trumpp A, manuscript in preparation), growth, and tumorigenesis.13, 14 c-Myc is a transcription factor that activates or represses two sets of target genes, which are thought to mediate its large variety of biological functions, including proliferation, cellular growth, apoptosis, angiogenesis, cell adhesion, and differentiation.15–17 In liver, the transcription of c-myc has been shown to be strongly induced in response to partial hepatectomy, and its overexpression leads to an increase in hepatocyte proliferation and size, which can eventually lead to the formation of liver cancer.13, 14 Nevertheless, the physiological role of c-Myc during liver homeostasis, as well as in hyperplastic responses, remains poorly understood. Here we generated mice lacking c-Myc in the adult liver and use these to address its role during TCPOBOP-induced direct hyperplasia.

Abbreviations

BrdU, bromodeoxyuridine; CAR, constitutive androstane receptor; ChIP, chromatin immunoprecipitation; FoxM1, forkhead transcription factor; Mcm, minichromosome maintenance complex; mRNA, messenger RNA; PCR, polymerase chain reaction; RT, reverse transcription; SD, standard deviation; TCPOBOP, 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene.

Materials and Methods

Generation of c-Myc–Deficient Mice.

The c-mycflox/flox mice18 were crossed with MxCre19 or HNF4αCre20 transgenic mice to obtain animals in which c-myc was either inducibly (MxCre; c-mycflox/flox) or constitutively (HNF4αCre; c-mycflox/flox) deleted. Interferon alpha induced deletion of c-myc was achieved in the MxCre mice by five intraperitoneal injections of polyI-polyC (Invivogen) once every 2 days, as described.21 PolyI-polyC injected c-mycflox/flox littermates served as controls. Southern blot analysis, Taqman polymerase chain reaction (PCR), and real-time reverse transcription (RT) PCR on DNA or RNA isolated from total liver was performed as previously described.18, 21

Primers used for RT-PCR are listed in supplement 6. All animal studies were conducted under federal guidelines and were approved by the Bundesamt für Veterinärwesen Authorization No.VD1729.

Histology, Immunohistochemistry, and BrdU Labeling.

All tissues were fixed in 10% formalin or 4% paraformaldehyde, paraffin embedded, and sectioned at 3 to 6 μm for hematoxylin-eosin staining or immunostaining procedures as previously described.21 The following antibodies were used: mouse anti-Ki67 (1:100, Novocastra), mouse anti-bromodeoxyuridine (BrdU) (1:500, Sigma), rabbit anti–c-Myc (1:500, Upstate Biotechnology), and rabbit anti–N-myc (1:200, Santa Cruz). For BrdU labeling, mice were first injected intraperitoneally 90 minutes before corn oil or TCPOBOP treatment with 100 μg BrdU (Sigma) per gram of body weight and were then maintained on BrdU in their drinking water (0.8 mg/mL) to achieve continuous labeling until mice were sacrificed.

Cell Size Assays.

A random field was defined for counting hematoxylin-eosin–stained liver sections (20× lens). At least six random fields per slide and three slides per liver sample were counted. Cells were quantified manually in each field and the mean ± standard deviation (SD) was determined for each mouse.

TCPOBOP Treatment.

TCPOBOP (0.6 mg/mL dissolved in corn oil) was injected intraperitoneally at a dose of 3 mg/kg body weight. After 3 days, the mice were sacrificed, and the total body and liver weight was measured. The controls were injected with an equal volume of corn oil. Each group contained four mice, and the experiment was repeated twice.

Microarray Analysis.

Microarray analysis was performed on MOE430v2 chips (Affymetrix) using RNA isolated from control and mutant littermates, 0 and 33 hours after TCPOBOP treatment. Microarray results for selected targets were confirmed using RT-PCR. Software used for analysis included the following: Remote Analysis Computation for Gene Expression,22 Database for Annotation, Visualization, and Discovery,23 Kyoto Encyclopedia of Genes and Genomes database,24 as well as the commercially available programs, GeneSpring (Agilent) and Ingenuity Pathway Analysis (Ingenuity Systems). For details see Supplement 6. Original array data can be accessed at the public repository ArrayExpress maintained by the European Bioinformatics Institute.

Chromatin Immunoprecipitation.

Three normal and c-Myc–deficient livers were isolated and chromatin immunoprecipitation (ChIP) was performed as described,25 details in Supplement 6. The antibodies used at 2 mg/sample were c-Myc N-262 (Santa Cruz), N-Myc C-19 (Santa Cruz), Histone H3 trimethyl-Lysine4 (Abcam), and Keratin10 (Covance). Analysis was performed using Sybr Green mix and the light cycler (Roche). Results are expressed as fold enrichment of the specific DNA sequence in the ChIP with the c-Myc antibody relative to the Keratin10 antibody used as negative control. For practical reasons, HNF4aCre; c-mycflox/flox mice were used, which have a TCPOBOP response indistinguishable from MxCre; c-mycflox/flox mice (data not shown).

Quantitative RT-PCR.

To analyze messenger RNA (mRNA) levels, RNA was isolated from homogenized livers using TRIzol (Invitrogen). Reverse transcription was performed using the StrataScript kit (Stratagene). Real-time PCR for relative quantification was done with Sybr Green on a light cycler (Roche). Each measurement was done in duplicate.

Western Blot.

Western blot was carried out as described in Supplement S6. Antibodies used were cyclin A (C-19, 1:100, Santa Cruz) and β-actin (1:100, Santa Cruz).

Results

Liver Function Is Maintained After Acute Loss of c-Myc.

To determine the role of c-Myc in the adult liver, we induced deletion of c-myc in MxCre; c-mycflox/flox mice using polyI-polyC treatment to induce interferon-alpha activity (Fig. 1A). Efficiency of recombination of the floxed c-myc allele was monitored by southern blot and real-time PCR (Fig. 1B and data not shown). Quantification revealed a 95% ± 4% deletion frequency at the c-mycflox locus (Fig. 1C). Real-time RT-PCR quantification of c-myc mRNA confirmed the elimination of c-myc transcripts in mutant livers (data not shown).

Figure 1.

MxCre-mediated deletion of the c-mycflox allele in the liver. (A) Diagram of the MxCre, c-mycflox, and c-mycDORFrec alleles; loxP sites are represented by red triangles, orange ovals represent the probe used for southern blot. (B) Southern blot of liver DNA obtained from c-mycflox/flox (cont) or MxCre; c-mycflox/flox (mut) mice 18 days after induction of deletion. The probe hybridizes to the c-mycflox allele (1926 bp) and to the recombined c-mycΔORFrec allele (1170 bp). (C) Quantification of MxCre-mediated recombination efficiency at the c-mycflox locus, mean (SD, n = 6). (D) Hematoxylin-eosin sections of both control and mutant livers; inset shows higher magnification of the same section.

Mice inducibly deleted for c-myc in the liver were normal and did not become overtly ill after MxCre induction. Nevertheless, histological analysis showed an altered hepatic morphology in experimental mice (Fig. 1D). Surprisingly, transcriptional analysis of livers from both c-myc–deleted mice and control littermates showed that the deletion of c-myc led to changes in only a small number of genes (212 genes significantly changed twofold or more), among them the up-regulation of numerous heat shock genes and genes involved in steroid biosynthesis (Supplement 1). Among genes implicated in the TCPOBOP response, the cytochrome p450 reductase important for reducing the xenobiotic-CYP complex, an important step in detoxification, was down-regulated (2.37× ↓). Cytochrome p450 reductase induction occurs on xenobiotic treatment and is partially dependent on CAR.26 Cyps require a heme group to oxidize substrates. Interestingly, Alas1, which catalyzes the rate-limiting step in heme biosyntheses and is responsive to TCPOBOP in a CAR-independent manner,26 is up-regulated in c-Myc–deficient mice (2.55× ↑). Another CAR target, Sult1e1,4 a sulfotransferase, was down-regulated (4.39× ↓). Two adenosine triphosphate–binding cassette (ABC) transporters normally up-regulated in response to TCPOBOP27 are also down-regulated, Abcb1a (2.53× ↓) and Abcc4 (Mrf4) (2.97× ↓). Thus it appears that components of the CAR pathway are compromised on deletion of c-myc during homeostasis. These changes, however, do not significantly impair the metabolic function of the liver during homeostasis (Supplement 2).

c-Myc Is a Mediator of TCPOBOP-CAR–Induced Hepatomegaly.

TCPOBOP is known to induce acute hepatocyte proliferation and growth (hepatomegaly) because of a combination of hypertrophy (increase in cell size and cell growth) and hyperplasia (increase in cellular division and cell number).4, 8 Because c-Myc has been implicated in both of these processes,16, 17 we examined whether its expression is induced in response to TCPOBOP. Indeed, c-myc mRNA was induced as soon as 6 hours after treatment and was maintained for at least 3 days (Fig. 2A). Strikingly, this induction was not observed in livers lacking the CAR gene, suggesting that c-myc may be a downstream target gene of CAR (Fig. 2A). Because c-Myc is also regulated posttranscriptionally,16 we analyzed protein expression in TCPOBOP-treated livers. As expected, nuclear c-Myc was induced in livers of TCPOBOP treated control but not in c-Myc–deficient livers (Fig. 2B), identifying c-Myc as a target downstream of the TCPOBOP-CAR signaling pathway.

Figure 2.

c-Myc is downstream of CAR and required for a full proliferative response upon TCPOBOP treatment. (A) Northern analysis of c-myc RNA expression kinetics in both wild-type (CAR+/+) and CAR-deficient (CAR−/−) mice after TCPOBOP administration. (B) Immunohistochemical analysis of c-Myc expression in c-mycflox/flox (control) or MxCre; c-mycflox/flox (mutant) livers 3 days post-TCPOBOP treatment. (C) Liver/body mass ratio, in both control and mutant animals, before and 3 days after treatment with TCPOBOP. Results are mean of 10 mice per group (SD). (D) Expression of Ki67 and incorporation of BrdU assayed by immunohistochemistry on liver sections in control or mutant liver after 3 days of TCPOBOP treatment. The graphs show the quantification of Ki67pos and BrdUpos cells per field. Results are the mean of three mice and four random fields were counted per mouse (SD). *** = Statistically significant, (Ki67, P = 0.0037, BrdU, P = 0.0075) (E) Analysis of liver cell size, 3 days after administration of TCPOBOP or corn oil (solvent). Quantification of relative cell size per field of individual control and mutant mice (n = 4). Results are the mean of four random fields counted (SD).

To determine whether c-Myc is required for TCPOBOP-induced hepatomegaly, control and c-Myc mutant mice were treated with one dose of TCPOBOP. As expected, control mice almost doubled their liver mass within 3 days (Fig. 2C). Mutant livers also responded but in an attenuated fashion, with liver size increasing by 55% (n = 10) compared with treated control littermates (100%), demonstrating a partial requirement for c-Myc in the TCPOBOP response. To dissect whether the hypertrophic or hyperplastic response is c-Myc dependent, we first examined the proliferation status of hepatocytes by assaying for the expression of both the cell cycle marker, Ki67, and BrdU incorporation assays. Although 71% of liver cells showed Ki67 expression in TCPOBOP-treated control livers (Fig. 2D), this value was decreased to 45% in mutants. Similarly, only 33% of mutant hepatocytes entered S phase, as determined by BrdU, versus 50% of control hepatocytes during the first 3 days post-TCPOBOP treatment (Fig. 2D). To evaluate hypertrophy, we examined hepatocyte size in control and mutant mice after TCPOBOP treatment. This analysis showed a similar increase of hepatocyte size (approximately 12%) in both control and mutant animals (Fig. 2E), suggesting that c-Myc is dispensable for TCPOBOP-induced hepatocyte hypertrophy. Thus, although the growth response was unaltered, the proliferative response was significantly inhibited but importantly not completely blocked. These results show that even though c-myc is one critical target of CAR, other still elusive CAR-dependent genes contribute to achieve a full proliferative response to TCPOBOP.

Profiling of the TCPOBOP-Induced Transcriptional Response in the Liver Identifies FOXM1 as an Important Downstream Effector.

To determine the basis of TCPOBOP-induced hepatomegaly, RNA from control livers before and 33 hours after TCPOBOP treatment was isolated and examined using microarrays. TCPOBOP treatment caused 1129 genes to significantly change their expression by more than twofold (Supplement 3). RT-PCR analysis on 10 selected genes confirmed the results obtained by the microarray analysis (Fig. 3C). Known downstream target genes of CAR signaling, the cytochrome p450 proteins, Cyp2b10 (9.47× ↑) and Cyp2c55 (85.4× ↑) were strongly up-regulated. We also observed the down-regulation of Sult1b1 (1.73× ↓) and the up-regulation of the ABC transporters, Abcc2 (2.45× ↑), Abcc3 (2.12× ↑), and Abcc4 (2.67× ↑). Our data also confirmed the observation28 that CAR affects liver metabolism through suppression of hepatic nuclear factor-4 activity (1.76× ↓), which normally drives the expression of Pck1 (4.1× ↓, important in gluconeogenesis) and Cyp7a1 (6.71× ↓, important in bile acid synthesis). In summary, these data confirm that 33 hours after TCPOBOP treatment, CAR signaling is highly active, as is reflected by high levels of expression of several CYP proteins and ABC transporters, whereas in contrast various aspects of liver metabolism are attenuated.

Figure 3.

Transcriptional Analysis of the TCPOBOP response in control and mutant livers. (A) Cell cycle pathway from the Kyoto Encyclopedia of Genes and Genomes database demonstrating how gene expression is altered 33 hours after TCPOBOP treatment. Each box represents a gene and the fill color of the box represents the fold expression change of that particular gene in response to TCPOBOP treatment in control livers. Boxes that are outlined in red identify those genes whose response to TCPOBOP is compromised in c-Myc deficient livers. The red numbers next to the red boxes indicate the fold down-regulation in the c-Myc deficient livers versus control after TCPOBOP treatment. (B) Graph demonstrating the levels of induction or suppression of various genes in response to TCPOBOP treatment at 33 hours as observed on the microarray. (C) Real-time RT-PCR analysis of selected genes confirming the results obtained by the microarray analysis.

The most significantly changed pathway after 33 hours of TCPOBOP treatment was the “cell cycle,” with a marked up-regulation of genes regulating the G1/S-transition, cyclinA2, cyclinD1, Gadd45b, and proliferating cell nuclear antigen (Fig. 3A, B). Unexpectedly, expression of genes known to inhibit the cell cycle such as p21CIP1, p107, and p18INK4c were also found to be up-regulated, which may indicate putative negative feedback loops. Licensing of replication forks late in G1 phase requires the minichromosome maintenance proteins (Mcm 2-7).29 Four of the six Mcm genes (Mcm 2, 5, 6, and 7) were significantly up-regulated in response to TCPOBOP (Fig. 3A, B and Supplement 3). In addition, Dbf4 (activator of S phase kinase) expression, a regulatory component of the Dbf4/Cdc7 kinase, which activates the Mcm 2-7 complex,30 was also increased in response to TCPOBOP (Fig. 3B).

The transition from G2 into mitosis strictly depends on the activation of cyclinB/Cdk1,31 which showed a striking increase on TCPOBOP treatment (Fig. 3A-C). A more than 50-fold increase was also observed in the G2-cyclin, cyclinA2. Moreover, expression of polo-like kinase-1 implicated in the activation of cyclinB/Cdk132 was 12-fold higher in TCPOBOP-treated livers compared with untreated controls. Several genes (Bub1, BubR1, and Mad) encoding proteins involved in the formation of the spindle assembly checkpoint, as well as Cdc20 (up-regulated over 50-fold) known to control the onset of anaphase via regulation of the anaphase-promoting complex,33, 34 were strongly up-regulated (Fig. 3A-C). In summary, these data suggest that TCPOBOP leads to the transcriptional activation of a significant number of genes known to be crucial for DNA replication, the G2-M transition, as well as controlling the spindle assembly checkpoint and the onset of mitosis.

The TCPOBOP-induced cell cycle program that we have uncovered, interestingly, overlaps with one recently reported to be controlled both in fibroblasts and the human osteosarcoma epithelial cell line, U2OS, by the forkhead box protein, FoxM1.7, 11 In liver, this protein has been shown to be important during liver regeneration,7, 35 and indeed FoxM1 transcription was up-regulated in response to TCPOBOP (Fig. 3A-C; Supplement 3), confirming results reported by others.10 Overall, our dataset shows that TCPOBOP not only induces CAR target genes, but also genes regulating all phases of the cell cycle. Because these data significantly overlap with previously reported transcriptional analyses of TCPOBOP-treated livers using smaller microarrays, TCPOBOP apparently induces a very robust and highly reproducible transcriptional response in livers.6, 36

Finally, the up-regulation of the S/G2 phase cyclinA protein on TCPOBOP treatment was attenuated in c-Myc–deficient livers, suggesting a compromised progression of the cell cycle (Supplement 7).

Metabolic Target Genes of CAR Are Not Dependent on c-Myc.

To determine how c-Myc deficiency affects the metabolic target genes of CAR, we examined the transcriptional profiles of livers from control and c-Myc deficient mice, 33 hours after the initiation of TCPOBOP treatment. As expected, mutant livers show a significantly altered transcriptional signature in response to TCPOBOP treatment (Supplement 5). As described previously, the CAR response pathway appears to be somewhat compromised in the homeostatic state in c-Myc deficient livers. We therefore examined the response of the CAR signaling pathway 33 hours after TCPOBOP induction in c-Myc mutant livers. Although CAR mRNA levels are lower in mutant livers during homeostasis, in response to TCPOBOP treatment we observe various Cyp proteins, Cyp2b10 (15.89× ↑), Cyp3a41 (4.58× ↑), and Cyp3a13 (3.15× ↑), which are known targets of CAR to be strongly up-regulated even in the absence of c-Myc.

Similarly, the cytochrome p450 reductase protein, which is expressed at low levels in mutant livers during homeostasis and is important for the functional activity of Cyp proteins, is up-regulated in response to TCPOBOP treatment (3.08× ↑). Other CAR targets down-regulated in the c-Myc deficient livers, such as the ABC transporters, Abcb1a (1.37× ↑), Abcc4 (Mrf4) (6.62× ↑), as well as the sulfotransferase, Sult1e1 (16.44× ↑), are dramatically up-regulated. In conclusion, the absence of c-Myc during TCPOBOP treatment does not significantly affect the initiation of CAR signaling itself, nor the transcription of many of CAR's metabolic targets.

TCPOBOP-Mediated Activation of FoxM1 and Cell Cycle Genes Require c-Myc.

To determine how the absence of c-Myc might compromise the TCPOBOP-induced proliferative response (Fig. 2D, Supplement 7), we examined the differences between control and c-Myc deficient livers after TCPOBOP treatment. Our analysis showed 265 genes that change twofold or more in treated mutants compared with control animals. Most strikingly, the most overrepresented pathway was again the “cell cycle” (Supplement 4). Close examination of the genes in this group showed that the TCPOBOP induction of most cell cycle genes described was severely attenuated in c-Myc deficient livers. For example, cyclin A2/B1/B2/B4 and Cdk1 all decreased several fold in mutant livers, and Cdc20 was reduced by more than 10-fold (Fig. 3A-C and Supplement 5). In addition, three of the six MCM genes and Dbf4 decreased in mutant livers. Importantly, FoxM1, the potential up-stream regulator of these genes, failed to be significantly induced by TCPOBOP in the absence of c-Myc activity (Fig. 3A-C and Supplement 5). This raised the hypothesis that c-Myc dependent FoxM1 activation could be responsible for the TCPOBOP response. Although FoxM1 transcription has been postulated as a candidate mediating the hyperproliferative response to TCPOBOP,7, 10 the mechanism by which CAR might be activating FoxM1 has remained unclear.

FoxM1 Is a Direct Target of c-Myc.

In a large-scale chromatin immunoprecipitation study using human cell lines, it has recently been reported that c-Myc binds to an E-box in the human FoxM1 gene.37 This E-box is also conserved in the mouse FoxM1 promoter (Supplement 6), suggesting that FoxM1 expression might be regulated by c-Myc. To determine whether c-Myc binds to the FoxM1 promoter in response to TCPOBOP treatment, chromatin immunoprecipitation (ChIP) experiments were performed. As is shown in Fig. 4A, a significant 3.4-fold enrichment of c-Myc binding to the FoxM1 E-box was found on TCPOBOP activation. Because untreated control and c-Myc deficient livers were indistinguishable, these data suggest that binding of c-Myc to the FoxM1 promoter occurs in a TCPOBOP-dependent manner. In contrast to c-Myc, no apparent binding of N-Myc to the FoxM1 promoter was detectable neither in response to TCPOBOP treatment nor in livers lacking c-Myc with or without TCPOBOP activation (Fig. 4B). Consistently, N-Myc mRNA levels were not elevated in c-Myc deficient mice treated with TCPOBOP compared with TCPOBOP-treated control mice, as measured by quantitative RT-PCR (data not shown). Together, these results demonstrate that FoxM1 is a direct transcriptional target of c-Myc and that the binding of c-Myc to the FoxM1 promoter only occurs in response to TCPOBOP treatment.

Figure 4.

c-Myc directly binds to the FoxM1 promoter in response to TCPOBOP. (A) Binding of c-Myc to the E-Box in the FoxM1 promoter region. ChIP was performed with either c-Myc or control antibodies from sonicated liver samples from either corn oil or TCPOBOP-treated, control, or c-Myc–deficient livers. **Significantly changed (P = 0.014). (B) ChIP analysis as in (A), but for N-Myc. No binding of N-Myc to the E-Box in the FoxM1 promoter region is observed. (C) Model for the role of c-Myc in TCPOBOP/CAR signaling. (see text for details).

In summary, most TCPOBOP-induced cell cycle genes depend on the activity of c-Myc. However, most genes are still up-regulated to some degree even in the absence of c-Myc (Figs. 3B, 3C, 4C, Supplement 5), suggesting that additional factors contribute to the transcriptional regulation of these genes. These findings are consistent with our observations that some c-Myc deficient liver cells are able to still proliferate in response to TCPOBOP.

Discussion

Despite the dramatic effects of xenobiotics such as TCPOBOP on the liver, the downstream signaling pathways initiated remain largely unknown. The detoxification of xenobiotics is one of the essential functions that the liver accomplishes by signaling through nuclear receptors. These nuclear receptors up-regulate cytochrome P450 family proteins, other drug-metabolizing enzymes, as well as transporters, which ultimately results in detoxification. Xenobiotics, such as TCPOBOP, that activate the nuclear receptor, CAR, also promote hepatomegaly, leading to an increase in both hepatocyte growth and proliferation, which can result in hepatocarcinogenesis.5

Our results demonstrate that the proliferative response to TCPOBOP is to a significant level dependent on c-Myc. Treatment of the liver with TCPOBOP induces c-Myc expression and this induction is blocked in mice that lack CAR, thus placing c-Myc downstream of TCPOBOP and CAR. It seems likely that CAR directly regulates c-Myc transcription, although we have been unable to identify a functional CAR binding site in the c-Myc proximal promoter. Furthermore, livers lacking c-Myc have a significantly attenuated TCPOBOP-induced proliferative response (Fig. 2D). This is consistent with the partial loss of TCPOBOP-induced growth and proliferation in mice lacking Mdm28 and indicates that c-Myc is one of several key factors required for this response. It is interesting to note that even in the absence of c-Myc proliferation still occurs, albeit at a lower level, suggesting that at least in vivo c-Myc may only be essential for proliferation in response to some but not other mitogenic signals. This is consistent with observations in other cell types in which c-Myc–independent proliferation occurs.12, 21, 38

Transcriptional profiling of the TCPOBOP-treated liver revealed the induction of a specific set of cell cycle genes whose expression was severely attenuated in the absence of c-Myc. The normal TCPOBOP-induced cell cycle program includes some genes involved in G1 control, but also numerous genes controlling replication and the G2/M transition. Interestingly, this set of genes strikingly overlaps with the genes previously shown in fibroblasts and U2OS cells to be controlled by the FoxM1 transcription factor.11 Indeed, FoxM1 has been previously reported as being up-regulated in response to TCPOBOP,10 and FoxM1-deficient hepatocytes show defects in replication and mitosis during liver regeneration35 and are resistant to carcinogen-induced hepatocellular carcinoma.7, 39 FoxM1 is also induced in response to bile acid activation of the nuclear receptor FXR (Farnesoid X receptor), and its induction in response to partial hepatectomy is completely absent in FXR-deficient mice, which also show attenuated induction of c-myc and defective liver regeneration.40

FoxM1 transcription has been postulated to be a mediator of the hyperproliferative response to TCPOBOP, although this has not yet been experimentally addressed.7, 10 In addition, the mechanism by which TCPOBOP-CAR might be activating FoxM1 was unknown. The microarray data confirmed by quantitative RT-PCR presented here demonstrates that the transcription of FoxM1 in response to TCPOBOP is c-Myc dependent. This observation is strongly confirmed by our ChIP experiments demonstrating that c-Myc binds to a conserved E-box in the promoter of FoxM1 on TCPOBOP treatment. Furthermore, FoxM1 may in turn also regulate c-Myc in a positive feedback loop, as it has recently been reported that FoxM1 transactivates the human c-Myc promoter synergistically with the trans-acting transcription factor 1, Sp1, 1 via binding to TATA-boxes.41, 42 Because these sequence motifs are evolutionarily conserved between mice and humans, transactivation of c-Myc by FoxM1 could initiate a feed-forward loop. This could not only amplify but also maintain the expression of c-Myc/FoxM1 target genes even after a xenobiotic stimulus has been cleared from the liver.

However, FoxM1 is probably not the only c-Myc target gene, because many of the cell cycle genes downstream of FoxM1 have also been shown to be direct targets of c-Myc itself. Examples include cyclinA2; cyclinB1; CDK1; MCM-3, MCM-5, MCM-6, MCM-7, and FoxM1 (http://genomebiology.com/2003/4/10/R69)37, 43, 44 (Fig. 4C). This raises the possibility that, after the activation of FoxM1 by c-Myc, both proteins may cooperate in the induction of several common target genes, a hypothesis future promoter studies should be able to address.

In summary, these results demonstrate a central function for c-Myc as a key component in the TCOPOBOP-induced CAR-c-Myc-FoxM1 signaling pathway that promotes hepatocyte proliferation. A better understanding of the mechanisms that regulate FoxM1 may eventually lead to the development of therapeutics that could promote increased hepatocyte proliferation. This would in particular be important in older patients undergoing liver transplantation. Indeed, it has been recently shown that transplanted hepatocytes overexpressing FoxM1 more efficiently repopulate the injured liver.45 Finally, because c-Myc is not essential for normal liver homeostasis, these data raise the possibility that specific inhibition of the c-Myc pathway may be effective to target liver cancer without affecting normal liver function and maintenance.

Acknowledgements

The authors thank Dr. Ingrid Grummt for probes, Giani-Filippo Mancini as well as other members of the ISREC MIM facility for help with histology, Dr. Keith Harshman, Dr. Otto Hagenbüchle, and Sophie Wicker, as well as other members of the Lausanne DNA Array Facility (DAFL) for excellent service. We also would like to thank all members of the Trumpp laboratory for helpful discussions.

Ancillary