| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
* Department of General, Visceral and Transplantation Surgery, Hannover Medical School, Hannover, Germany; and
Drug Research and Medical Biotechnology, Fraunhofer Institute of Toxicology and Experimental Medicine, Hannover, Germany
1Correspondence: Drug Research and Medical Biotechnology, Fraunhofer Institute of Toxicology and Experimental Medicine, Nikolai-Fuchs-Strasse 1, 30625 Hannover, Germany. E-mail: borlak{at}item.fraunhofer.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
were selectively regulated in liver metastases. We therefore studied protein expression of regulated transcription factors and found unacetylated HNF6 to be a hallmark of colorectal liver metastases. For its known interaction with HNF6, we investigated expression of FOXA2, which we found to be specifically induced in colorectal liver metastases. By electromobility shift assay, we examined DNA binding of disease regulated transcription factors. Essentially, no HNF6 DNA binding was observed. We also searched for sequence variations in the DNA binding domains of HNF6, but did not identify any mutation. Furthermore, we probed for expression of 28 genes targeted by HNF6. Mostly transcript expression was repressed except for tumor growth. In conclusion, we show HNF6 protein expression to be driven by the hepatic environment. Its expression is not observed in healthy colon or primary colonic cancer. HNF6 DNA binding is selectively abrogated through lack of post-translational modification and interaction with FOXA2. Targeting of FOXA2 and HNF6 may therefore enable mechanism-based therapy for colorectal liver metastases.—Lehner, F., Kulik, U., Klempnauer, J., Borlak, J. The hepatocyte nuclear factor 6 (HNF6) and FOXA2 are key regulators in colorectal liver metastases.
Key Words: liver-enriched transcription factors • colon cancer • human liver • hepatocellular carcinoma
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. A frequent complication of this malignancy is the development of colorectal liver metastases, which results in impaired liver function and eventually in hepatic failure (2)
. Despite improved methods for detection and treatment, the patients’ overall 5-year survival remains poor, ranging from 24 to 44% (3)
. Research is needed to enhance our understanding of the molecular causes of colorectal liver metastases and to identify novel opportunities to treat this secondary liver malignancy.
Several investigators have reported that altered expression of liver-enriched transcription factors affects growth and differentiation of liver cells in hepatocellular carcinomas (HCC) (4
5
6)
. There is evidence for loss of activity of the liver-enriched transcription factor HNF4
to result in progressive HCC (7)
. Forced expression of HNF4 promoted reversion of tumors toward a less invasive, highly differentiated, slow-growing phenotype. This points to a pivotal role of liver-enriched transcription factors in disease. Essentially, transcription factors are master regulatory proteins that interact with many different molecules, including coactivators, repressors, enzymes, DNA, and RNA, to control gene expression. Such interactions will inevitably repress or activate gene expression and therefore determine cellular phenotype. Indeed, numerous studies have established the pivotal role of liver-enriched transcription factors in organ development and liver function. There is conclusive evidence for transcription factors to act in concert and to enable cellular differentiation and regulation of metabolic functions (8
9
10
11)
. Some of these transcription factors are tissue-specific; others are required for gene expression in a variety of tissues. In general, liver-enriched transcription factors are classified by their ability to recognize specific DNA binding motifs and are divided into major families (e.g., hepatocyte nuclear factors 1, 3, 4, and 6 and their isoforms as well as the CAAT/enhancer binding proteins, including the many subfamily members, as reviewed elsewhere; refs. 12
, 13
).
It is of considerable importance that gene expression profiling of human hepatocellular carcinoma revealed a role of liver-enriched transcription factors in HCC (4
5
6)
. To the best of our knowledge, expression and activity of these proteins in human colorectal liver metastases and in primary tumors of the colon have not been studied. Because liver-enriched transcription factors are involved in multiple regulatory pathways, changes in the expression and activity of these proteins will likely affect the transcriptional network of genes targeted by these factors, thereby determining cellular phenotype and biological behavior of tumor cells.
Here we report our search for disease-associated changes of liver-enriched transcription factors and of genes targeted by these factors in the healthy human liver, in the colon, in primary colonic tumors, and in colorectal liver metastases. Unlike healthy colon and/or primary colonic tumors, we found HNF6 to be abundantly expressed in colorectal liver metastases as a result of growth in a hepatic environment. This demonstrates considerable plasticity of metastatic tumor cells. Despite its abundant expression, HNF6 was unable to bind to known recognition sequences of genes targeted by this factor. A major finding of our study was the tumor-specific induction of FOXA2. This protein is known to interact with HNF6 and functions as a transcriptional repressor for HNF6-targeted genes. We observed a lack of post-translational modification of HNF6 to render the protein unable to bind DNA. Targeting FOXA2 will likely restore HNF6 activity, thereby interfering with tumor growth.
Overall, this study aimed to identify culprit transcription factors and genes targeted by these factors in primary colonic tumors and colorectal liver metastases as a mechanism of disease.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
provides a synoptic overview of patients’ characteristics including age, gender, primary diagnosis, and tumor staging as well as time span between primary tumor and colorectal liver metastases. Notably, the patients reported in Table 1
received surgery of either the colon (e.g., primary tumor resection) or colorectal liver metastases, but none of the patients received surgery for colon and liver tumors simultaneously.
|
Explanted human material
The primary colonic tumors and/or the colorectal liver metastases were removed by standard surgical procedures. All surgical specimens were subjected to histopathology. Excised healthy and tumorous tissue was shock-frozen in liquid nitrogen and stored at –80°C until analyzed.
RNA isolation and cDNA synthesis
RNA was isolated form tissue samples using the RNeasy Mini Kit (Quiagen, Valencia, CA, USA) according to the manufacturer’s recommendation. Quality and quantity of isolated RNA were checked by capillary electrophoresis (Bioanalyzer 2100, Agilent Technologies, Santa Clara, CA, USA) following the manufacturer’s instructions or by gel electrophoresis. Two micrograms of total RNA from each sample were used for reverse transcription (RT). RNA and random primer (Promega, Mannheim, Germany) were preheated for 10 min at 70°C, then chilled on ice for 2 min. A total of 5 x RT-avian myoblastosis virus (AMV) buffer (Promega), dNTP’s (10 mM), RNAsin, AMV-RT (all from Promega), and diethyl pyrocarbonate (DEPC) -H2O were added to a final volume of 20 µl. Reverse transcription was carried out for 60 min at 42°C and was stopped by heating to 95°C for 5 min. The resulting cDNA was frozen at –20°C until additional experimentation.
Thermocycler RT-polymerase chain reaction (RT-PCR)
Primer design was done with the program Primer 3 (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). Cross-reaction of primers with the genes was excluded by comparison of the sequence of interest with a database (Blast 2.2, U.S. National Centre for Biotechnology Information, Bethesda, MD, USA) and all primers used in our study were introne spanning. PCR reactions were undertaken with a 20 µl reaction mixture containing HotStarTaq Master Mix (Qiagen, Hilden, Germany), DEPC, 1 µl cDNA, and 1.0 µM concentration of the 3'- and 5'-specific oligomers (synthesized by Invitrogen, Hilten, Germany). PCR reactions were carried out on a thermal cycler (T3, Biometra, Goettingen, Germany). Detailed oligonucleotide sequence information and the PCR amplification protocol are given in Table 2
. DNA contamination was checked for by direct amplification of RNA extracts before conversion to cDNA. Contamination of RNA extracts with genomic DNA was assayed for by gel electrophoresis and by DNA digest prior to cDNA synthesis.
|
PCR reactions were done within the linear range of amplification, and amplification products were separated using 1.5% agarose gel and stained with ethidium bromide. Gels were photographed on a transilluminator (Kodak Image Station 440) and amplicons were quantified using the Kodak 1D 3.5 network software.
Quantitative PCR analysis of HNF6 and of the housekeeping gene mitochondrial ATPase with the Roche LightCycler® system
Real-time PCR was done with the LightCycler according to the manufacturer’s recommendations (Roche Diagnostics, Penzberg, Germany). RNA was prepared and reverse transcription was done as detailed above. We used SYBR® Green I as a fluorescent dye to determine the amplified PCR product after each cycle. The length of PCR products was checked by gel electrophoresis. Table 2
provides the oligonucleotide sequence for HNF6 and mitochondrial ATPase (e.g., the housekeeping gene). PCR was performed with 100 nM of HNF6 or MitATPase primers in a standard PCR reaction containing 50 ng of DNA, 4.0 mM MgCl2, and 2 µl of LightCycler DNA Master hybridization mixture (LightCycler DNA Master Hybridization Probes, Roche Diagnostics) in a total volume of 20 µl. The reaction for HNF6 was started with a denaturation step at 95°C for 20 s and amplification was performed for 50 cycles of denaturation (95°C for 0 s; ramp rate 20°C/s), annealing (58°C for 8 s, ramp rate 20°C/s), and extension (72°C for 18 s, ramp rate 20°C/s). In the case of the mitochondrial ATPase, the reaction was started with a denaturation step at 95°C for 20 s and amplification was performed for 50 cycles of denaturation (95°C for 0 s; ramp rate 20°C/s), annealing (55°C for 8 s, ramp rate 20°C/s), and extension (72°C for 18 s, ramp rate 20°C/s). PCR products were identified by monitoring DNA melting curves in the glass capillary. At the end of each extension phase, fluorescence was observed and used to quantify measurements within the linear range of amplification, yielding calculated concentrations as relative units. Exact quantification was achieved by serial dilution with cDNA produced from total RNA extracts using serial dilution steps. Values obtained for HNF6 were divided by those of mitochrondrial ATPase to obtain expression values relative to the housekeeping gene. Figure 1
depicts results obtained for HNF6 and mitochondrial ATPase by qRT-PCR (LightCycler) and the thermocycler method.
|
Western blot experiments
Western immunoblotting was done as follows: total protein (100 µg) or nuclear protein (30 µg) extracts from healthy or tumorous liver probes were denaturated at 95°C for 5 min, followed by sodium dodecyl sulfate PAGE (SDS-PAGE) on 12% polyacrylamide gels, and blotted onto a polyvinylidene difluoride membrane (NEN, Dreieich, Germany) at 350 mA for 2 h in a buffer containing 400 mM glycine and 50 mM Tris (pH 8.3). Nonspecific binding sites were blocked with Rotiblock (Roth, Karlsruhe, Germany) in 1x TBS buffer. After electroblotting of proteins, membranes were incubated with polyclonal antibodies for HNF6 (a kind gift from Dr. R. H. Costa, Chicago, IL, USA) for 1 h and washed three times with 1x TBS buffer containing 0.1% Tween-20 (Roth). Subsequently, membranes were incubated with a 1:5000 diluted anti -rabbit antibody (Chemicon, Hofheim, Germany) for 1 h at room temperature, followed by three successive washes with 1x TBS buffer containing 0.1% Tween-20 (Roth). Immunoreactive proteins were visualized with a chemiluminescence reagent kit (NEN) according to the manufacturer’s instructions, and bands were scanned with the Kodak Image Station CF 440 and analyzed using the Kodak 1D 3.5 imaging software (Eastman Kodak Company, Rochester, NY, USA).
Immunohistochemistry of HNF6 in human liver and colorectal liver metastases
Paraffin-embedded slices of human liver specimens (healthy liver and colorectal liver metastases) derived from patients P17, P18, P19, and P20 were kindly provided by Dr. M. Mengel, Department of Pathology, Medical School of Hannover. These sections were deparaffinized, demasked with 1M citratbuffer, incubated with 0.6% H2O2 in methanol for 30 min, and subsequently with protein block serum-free reagent (Dako, Glostrup, Denmark) for 8 min. Incubation with polyclonal antibody (Santa Cruz Biotechnology, Heidelberg, Germany) against HNF6 (sc 6559, 1:15 dilution) was performed for 45 min. The sections were rinsed with Tris-buffered saline, incubated with biotinylated universal secondary antibodies (Dako) for 15 min, and with horseradish peroxidase-conjugated streptavidin solution (Dako) for 15 min. Labeling was detected using a diaminobenzideine (Dako) for 5 min.
To confirm specificity of the immunohistochemical staining, images on the right panel of Fig. 8
show a tumor section of the same patient stained with the antibody preabsorbed for 2 h with a 5-fold excess of peptide antigen blocking the immunostaining.
|
Design of EMSA-oligos for HNF6 binding sites in promoter sequences of human genes
Known binding sites of HNF6 (ONECUT1) were collected from the TRANSFAC database, a database on gene regulation (www.biobase.de) that collects data on transcription factors and their binding sites in promoters and enhancers of eukaryotic genes. Our search was done with TRANSFAC release 9.4. To retrieve promoters of human genes, we used TRANSPro release 2.1. We used the HNF6 matrix (M00639; V$HNF6_Q6) and based our search on 13 known binding sequences for HNF6. We optimized the design for the oligo probes shown in Table 3
by studying promoter binding sites for HNF6 of mouse TTR (transthyretin) and HNF4 (hepatocyte NF 4) using sequences CTAAGTCAATAAT and aggatagAAGTCAATGAtctggga, respectively. We also interrogated the promoter binding sites of rat FOXA2 and PEPCK (phosphoenolpyrovat carboxykinase) using the sequences agcttaaggcccgATATTGATTTttttttctcc and caaagttTAGTCAATCAaacgttg.
|
Human binding sequences were estimated by alignment of the known sites with the promoter sequences of human orthologous genes. The human sequences with the best alignment were used for electrophoretic mobility shift assay (EMSA) studies as detailed below.
Preparation of nuclear extracts
For the preparation of nuclear extracts we used the protocol of Gorski et al. with modifications (14)
. After explantation, tissue specimens were placed on ice-cold PPB-containing buffer and swiftly transported to the laboratory (i.e., time of explantation to tissue preparation was 
30 min). Having determined the weights, tissue materials were cut into small pieces and placed into a vessel containing buffer HP1, which consisted of 25 x CompleteTM (Roche Diagnostics GmbH, Mannheim, Germany), 1 M DTT, 0.5 M EDTA (pH 8.0), 1 M ß-glycerophosphate, glycerine,1 M HEPES (pH 7.6), 1 M KCl, 200 mM Na3VO4, 0.1 M spermine, 3.44 M spermidine, and sucrose bidest H2O. Samples were homogenized with an Ultra-Turrax (Janke & Kunkel, IKA Labortechnik, Staufen, Germany). The suspension was then transferred to a 15 ml hand potter homogenizer (Wheaton, Millville, NJ, USA) to achieve a homogeneous suspension. Thereafter, the suspension was transferred into SW28 tubes (UltraClear Beckman, Palo Alto, CA, USA) and the volume was adjusted and centrifuged at 24,000 rpm at 2°C in a Beckman L7–55 ultracentrifuge for 1 h. After centrifugation, nuclear pellets were isolated and resuspended in buffer HP1 containing glycerine at a concentration of 19 volume percent. Again, the resuspended pellet was homogenized in a hand potter (Wheaton). Thereafter, the solution containing nuclear extracts was placed into SW28 tubes and centrifuged at 24,000 rpm at 2°C in a Beckman L7–55 centrifuge for 1 h. Finally, the supernatant was discarded and the pellet was suspended in 1–2 ml lysis buffer containing 25 x CompleteTM (Roche Diagnostics GmbH), 1 M DTT, 0.5 M EDTA (pH 8.0), 1 M ß-glycerophosphate, glycerine, 1 M HEPES (pH 7.6), 1 M KCl, 1 M MgCl2, 200 mM Na3HPO4, and bidest H2O. Again, this suspension was homogenized in a hand potter (Wheaton) and nuclei were inspected microscopically. DNA amount was determined spectrophotometrically by determining the ratio of optical densities 280 > 260 nm. The DNA concentration was adjusted to 0.5 mg DNA/ml suspension. Then 4M ammonium sulfate was added (1/10 of the final volume) and samples were placed on ice for 30 min. Thereafter, the solution was placed into a Ti70.1 tube (Beckman) and centrifuged at 40,000 rpm at 2°C for 1 h. The resultant supernatant was carefully removed and the volume was determined. Precipitation of nuclear extracts was achieved by addition of 0.3 g anhydrous ammonium sulfate per milligram supernatant and samples were placed on ice for 45–60 min. Finally, the ice-cold solutions were placed into Ti70.1 tubes (Beckman) and centrifuged at 40,000 rpm at 2°C for 1 h. The supernatant was then discarded and the pellet was taken up in a dialysis buffer containing 1 M DTT, 0.5 M EDTA (pH 8.0), glycerine, 1 M HEPES (pH 7.6), 1 M KCl, and bidest H2O. A final DNA concentration of 10 µg/ml was adjusted and samples were placed on ice for 30–60 min. Notably, these samples were placed into a Slide a Lyzer Dialysis cassette (Pierce, Rockford, IL, USA) and stored at 4°C. After 2 h of dialysis, the lysis buffer was replaced and samples were once again dialyzed for a further 2 h. The samples were taken from the cassette, placed into 1.5 ml Eppendorf vessels, and centrifuged at 14,000 rpm at 4°C for 5 min. The concentration of the resultant nuclear proteins was determined and the remaining nuclear proteins were stored at –80°C to await further analysis.
Annealing of synthetic oligonucleotides and [32P] labeling
Oligonucleotides representing a high-affinity consensus HNF6, NGN3, HNF3, HNF4, TTR, and PEPCK binding site were chosen. For sequence information, see Table 3
.
Oligonucleotides were annealed at a concentration of 19.2 pM · µl–1 in 200 mM Tris (pH 7.6), 100 mM MgCl2, and 500 mM NaCl at 80°C for 10 min, then cooled slowly to room temperature overnight and stored at 4°C. Annealed oligonucleotides were diluted to 1:10 in Tris-EDTA buffer (1 mM EDTA, 10 mM Tris, pH 8.0) and labeled using [32P] ATP (Amersham Biosciences Europe GmbH, Freiburg, Germany, 250 µCi, 3000 Ci · mM–1), and T4 polynucleotide kinase (New England Biolabs GmbH, Frankfurt am Main, Germany). End-labeled probes were separated from unincorporated [32P] ATP with a Microspin G-25 Column (Amersham Biosciences Europe GmbH, Freiburg, Germany) and eluted in a final volume of 100 µl.
EMSA
The procedure for EMSA was adapted from a previously described method (15)
. Briefly, 7.5 µg of nuclear extract was incubated with the binding buffer consisting of 25 mM HEPES (pH 7.6), 5 mM MgCl2, 34 mM KCl, 2 mM DTT, 2 mM Pefablock (Roche Diagnostics GmbH), 0.5 µl aprotinin (2.2 mg · ml–1, Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany), 50 ng poly (DL-dC), and 80 ng BSA (PAA Laboratories GmbH, Cölbe, Germany). The binding reaction was carried out for 20 min on ice, and free DNA and DNA-protein complexes were resolved on a 6% polyacrylamide gel. Competition studies were done by adding a 10-fold excess of unlabeled oligonucleotides to the reaction mix. For supershift studies, a specific HNF6 antibody (Santa Cruz Biotechnology, Inc., Heidelberg, Germany) was added to the reaction mix 10 min before addition of the labeled probe. Gels were blotted to Whatman 3 MM paper, dried under vacuum, exposed to imaging screens (Imaging Screen-K, Bio-Rad Laboratories GmbH, Munich, Germany) for autoradiography overnight at room temperature, and analyzed using a PhosphorImager (Molecular Imager FX pro plus; Bio-Rad Laboratories GmbH) and the Quantity One Version 4.2.2 software (Bio-Rad Laboratories GmbH).
Microarray experiments
Global gene expression analyses were done with n = 24 colorectal liver metastases and n = 10 healthy livers as detailed below:
RNA isolation and production of copy RNA
cRNA samples were prepared following the Affymetrix Gene Chip® Expression Analysis Technical Manual (Santa Clara, CA, USA). Briefly, total RNA was isolated from frozen tissue using QIAGEN’s RNeasy total isolation procedure. A second cleanup of isolated RNA was performed using the same RNA isolation kit. In all, 10 µg of total RNA was used for the synthesis of double-stranded cDNA with Superscript II RT and other reagents from Invitrogen Life Technologies. HPLC-purified T7-(dT)24 (GenSet SA) was used as a primer. After cleanup, double-stranded cDNA was used for synthesis of biotin-labeled cRNA (Enzo® BioArray High Yield RNA Transcript Labeling Kit, Affymetrix). cRNA purified with Rneasy spin columns from Qiagen was cleaved into fragments of 35–200 bases by metal-induced hydrolysis.
Array hybridization and scanning
A measure of 10 µg of biotinylated fragmented cRNA was hybridized onto the HG U95Av2 array, which contains 10,000 full-length genes.
The hybridized, washed, and colored arrays were scanned using the Agilent Gene Array® Scanner. Scanned image files were visually inspected for artifacts, then analyzed, each being scaled to an all-probe set intensity of 150 for comparison between chips. The Affymetrix Microarray Suite (version 5.0) was used to control the fluidics station, and the scanner to capture probe array data and analyze hybridization intensity data. Default parameters provided in the Affymetrix data analysis software were applied to run the analysis.
Data analysis
The hybridization values for each gene probe presented on the array with a set of 16 perfect and mismatch oligonucleotide pairs were calculated with the Affymetrix Microarry Suite 5.0 Software, using the manufacturer’s statistical algorithm. The results were reported as numeric expression values (signal intensities and absolute information), detection calls "present" or "absent," produced by two independent algorithms. The results of a single comparison analysis between two different arrays were reported for each gene as signal logarithm ratio (log2ratio) and a change called "increase" or "decrease." Multiple data from replicate samples were evaluated and compared using statistical analysis with the Affymetrix Data Mining Tool 3.0 (DMT). The average and SD statistics within Affymetrix DMT were used to summarize the expression level (signal values) for each transcript across the replicates. The unpaired one-sided t test converting P value to a two-sided P value was used to determine the level between sets of colorectal liver metastases and healthy liver tissue, with the P value cutoff determined as 0.05. Only those genes that were detected (had a "present" call) in all samples of colorectal liver metastases and healthy liver tissue were taken into consideration as differentially expressed. Fold-change values were calculated as the ratio of the average expression level for each gene between two tissue sets. Comparison ranking analysis was used to study the concordance of gene expression changes in pairwise comparisons of tumor samples with healthy liver tissue. The results are shown as % of "increase" or "decrease" calls in individual comparisons.
HNF6 DNA sequence analysis
DNA isolation
Genomic DNA from healthy human liver and colorectal liver metastases was isolated with the NucleoSpinTissue Kit (Macherey-Nagel, Düren, Germany) according to the manufacturer’s instructions. The quality and quantity of isolated genomic DNA were checked on ethidium bromide-stained 1% agarose gel using known lambda DNA concentration (Amersham, Freiburg, Germany) as standard.
PCR amplification
PCR primers were designed with the publicly available PRIMER3 (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi) software and published sequence information of HNF6 according to GenBank (NCBI) entry. The DNA binding domains (cut and homeodomain) were amplified (see Table 4
for the 3'- and 5'-specific primers) using oligonucleotide synthesized by Invitrogen. A standard PCR reaction consisted of 20 ng of genomic DNA, 2.5 µl of PCR buffer (10x), 0.2 µl of Taq polymerase (5 U/µl), 0.5 µl dNTPs (10 mM), and 0.5 µl of each primer pair (10 pmol/µl) adjusted to a volume of 25 µl with distilled water. Typical PCR conditions consisted of an initial denaturation of 95°C for 15 min, followed by 34 cycles of 94°C 10 s denaturation, 60°C 30 s annealing, 68°C 2 min elongation, and a final elongation at 68°C for 10 min. PCR reactions were carried out on Biometra thermocyclers. PCR products were analyzed on GelDoc 2000 (Bio-Rad) using ethidium bromide-stained 1% agarose gels and a 1 kb-plus ladder as size marker (Invitrogen). A negative control (water only) was included for PCR amplification.
|
Sequencing of DNA binding domains
Mutations were searched for double-stranded direct sequencing using gene-specific primers. Amplified fragments were purified with PCR clean-up kits according to the manufacturer’s protocol (QIAquick PCR Purification Kit, Qiagen), subjected to cycle sequencing with BigDyeTerminator v3.1 Kit following the manufacturer’s procedure, and injected to an ABI 3100 Genetic Analyzer (Applied Biosystems, Darmstadt, Germany). Sequences were analyzed for nucleotide changes using appropriate programs (SeqScape, Applied Biosystems). Sequence for HNF6 as published at GenBank (NCBI) was used as reference.
Hierarchical gene cluster analysis
Hierarchical gene cluster analysis was done according to Ward’s minimum variance algorithm. Gene expressions are given as signal intensities obtained from ethidium bromide-stained images. The genes are arranged as ordered by the clustering algorithm. The color image is proportional to transcript abundances, with dark green representing less abundant and red representing more abundant mRNA transcripts.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In our patient cohort of colon tumors (n=16), the median age was 64 years and the distribution of gender was 63% males and 37% females. Patients were diagnosed with primary colorectal cancer. This was confirmed by histopathology.
Preparation of tissue material from the liver or colon was such that biopsies were taken either within the tumor or within normal tissue, as detailed below.
Gene expression profiling of liver-enriched transcription factor in colorectal liver metastases
We studied gene expression of liver-enriched transcription factors in healthy liver and tumor tissues. Gene expression of transcription factors was computed relative to the housekeeping gene mitochondrial ATPase, which we found to be stably expressed (Fig. 2
). We determined the linear range of PCR amplifications for all liver-enriched transcription factors. Additional file 1 depicts the amplification products, which were separated on a 1.5% agarose gel and stained with ethidium bromide. All PCR reactions were done within the linear range. Expression of liver-enriched transcription factors did not differ statistically (Fig. 3
) except for HNF6, HNF1ß, and C/EBP. Furthermore, abundance of transcript expression of most transcription factors was less than that of mitochondrial ATPase. Expression of transcription factors varied among individual patients. Figure 4
depicts a representative ethidium bromide-stained RT-PCR gel for some patients. Expression of hepatic nuclear factors differed when healthy liver and colorectal liver metastases were compared, but did not reach statistical significance as observed for FOXA2, HNF4, HNF4, HNF4, CEBP, CEBPß, CDP, and GATA4.
|
|
|
We studied the expression of ALDH3A1, ADH1A1, COL5A1, CYP51, HSP105, and UGT1A1 in healthy liver and tumorous tissues (Fig. 3)
. These genes are bona fide targets of HNF6 (16)
. Based on computational analysis (Spearman’s correlation coefficient) we found expression of HNF6 and ADH1A1 and/or UGT1A1 to be significantly regulated in our patient cohort. Figure 5
depicts scatter blots for n = 29 patients. Note that the diagrams represent a relationship for either healthy or tumorous tissue. HNF6 also serves as a coactivator protein to enhance FOXA2 (=HNF3ß) transcription. At the gene expression level, no significant correlation between HNF6 and FOXA2 expression was obtained (scatter blots not shown).
|
We studied the expression of neurogenin 3 (NGN3), a bona fide target of HNF6 (17
, 18)
. NGN3 is a transcription factor required for specification of the endocrine lineage in uncommitted and multipotent intestinal progenitor cells. Mice homozygous for a null mutation in ngn3 fail to generate any intestinal endocrine cells or endocrine progenitor cells. There was considerable variation in the expression of NGN3 in healthy liver and colorectal liver metastases, with the mean expression of NGN3 being higher in tumor tissue. This suggested additional regulation of NGN3 in colorectal liver metastases, which appeared to be independent of HNF6.
Gene expression of HNF6 and of heat shock proteins HSP105B and HSP90 was significantly correlated in colorectal liver metastases. These genes are bona fide targets for HNF6 and act as chaperones in facilitating protein folding. Indeed, induction of proteosomal degradation of HSP90 superchaperone complexes is clinically perused to abrogate oncogenic protein expression (19)
. There was tight regulation between HNF6 and gene coding for collagen5A1 in healthy liver tissue, but less so in liver tumors.
Finally, we studied gene expression of CCAAT enhancer displacement protein (=CDP) and of GATA4 in healthy liver tissue and colorectal liver metastases. Notably, CDP may act as a competitive repressor for CCAAT protein-mediated transactivation of targeted genes (13)
and competes for binding sites of CEBPs in order to repress histone deacetylase activity. We found CDP gene expression to be statistically significantly induced in colorectal liver metastases (see box blot, Fig. 3
).
Furthermore, the GATA family of Zn-finger transcription factors participates in gastrointestinal development. Recent evidence suggests epigenetic silencing of GATA4 and GATA5 in colorectal and gastric cancer through promoter hypermethylation of CpG islands (20)
. We found GATA4 gene expression to be significantly reduced in colorectal liver metastases and were unable to amplify GATA5 transcripts in healthy or tumorous tissue.
We further investigated gene expression of insulin-like growth factor 1 (IGF1) in our patient cohort and found IGF1 mRNA levels to be significantly reduced in colorectal liver metastases compared with healthy liver tissue of the same patient.
Quantitative RT-PCR of HNF6 with the LightCycler system
As shown in Fig. 1
, results for HNF6 were comparable when the thermocycler and the LightCycler method for quantitative RT-PCR were used. Additional file 2 provides results of HNF6 gene expression for all individual patients.
Gene expression profiling in primary tumors of the colon
We studied the expression of liver-enriched transcription factors in tumor resection material of the colon. This enabled a better understanding of transcript regulation in colorectal liver metastases. Again, tissue material was qualified by histopathology and the expression of transcription factor was computed relative to the housekeeping gene mitochondrial ATPase, as described above. No statistically significant difference was obtained when transcript abundance in extracts of healthy colon tissue was compared with RNA extracts of tumor tissue. Strikingly, HNF6 was not expressed in healthy and/or tumor tissue (Fig. 6
). This contrasts our findings with liver resection material, where HNF6 transcripts were expressed, albeit at different levels. Notably, none of the transcription factors studied were significantly changed when healthy and tumorous colon were compared, even though expression of HNF1, HNF1ß, HNF4, HNF4, CEBP, PPAR, IgF1ß, AHR, and GATA4 differed among individual patients (Fig. 7
).
|
|
We used RNA extracts of healthy and tumorous colons to investigate expression of genes considered to be bona fide targets of HNF6. Essentially, we found ADH1A1 to be repressed whereas expression of the heat shock proteins HSP105B and HSP90 was increased. This was no surprise, as transcriptional activation of the aforementioned genes is not solely dependent on the proper function of HNF6. Exaggerated expression of heat shock proteins in tumor tissue is frequently observed and known to facilitate oncogenic protein expression. We studied the expression of NGN3. Its expression did not differ statistically when healthy tissue and tumor extracts were compared even though NGN3 is a bona fide target of HNF6.
Immunohistochemistry of HNF6 in healthy liver and colorectal liver metastases
Figure 8
depicts results for patients P17 to 20. The arrows in panel A exemplify HNF6 staining. HNF6 staining was more abundant in healthy liver parenchyma than with colorectal liver metastases. We blocked HNF6 binding with a specific peptide to demonstrate selectivity of the immunostains as detailed in Materials and Methods and shown in Fig. 8B
.
Western blot of HNF6 in healthy human liver and colorectal liver metastases
We investigated expression of HNF6 protein in nuclear extracts of healthy liver and colorectal liver metastases. As shown in Fig. 9
, our antibody detected two immunoreactive bands. We observed significant differences in the expression level of these bands when nuclear extracts of healthy liver and colorectal liver metastases were compared. With nuclear extracts of healthy humans, liver expression of the upper immunoreactive band was strong, but with nuclear extracts of colorectal liver metastases, a prominent lower immunoreactive band and a faint upper band were observed. The upper band corresponds to the acetylated form of HNF6; the lower band represents the nonacetylated form. For its known interaction with HNF6, we investigated the expression of FOXA2 (HNF3ß) in healthy liver and colorectal liver metastases. As shown in Fig. 10
A, expression of FOXA2 was significantly induced in colorectal liver metastases compared with healthy liver tissue. We also observed induction of HNF1ß in colorectal liver metastases (see Fig. 10B
). Our findings are highly suggestive for differences in the post-translational modification of the HNF6 protein in healthy and tumorous liver tissues. It has been reported that the stability of the HNF6 protein depended on acetylation by the CRB binding protein coactivator, with CBP acetylation of the HNF6 protein increasing its steady-state levels without influencing nuclear localization of HNF6 (21)
. It has also been shown that CBP acetylation influenced nuclear retention of other liver-enriched nuclear transcription factors. Expression of the HNF6 protein differed in healthy and tumorous liver tissues, with a low level of acetylated HNF6 in colorectal liver metastases (see Fig. 9
). Unfortunately, no commercial antibody is available to probe specifically for the acetylated HNF6 variant, but the study by Rausa et al. (21)
clearly demonstrates that transcriptional activity of the HNF6 protein depends on the CBP acetylation site in the cut domain of this protein. We studied by gene chip analysis the expression of genes targeted by HNF6, most of which we found to be significantly repressed compared with healthy liver tissue of the same patient. This agrees well with our finding of impaired HNF6 DNA binding in colorectal liver metastases, as will be discussed below. Notably, we did not detect HNF6 transcript in healthy colon or colonic tumor tissue. Hence, no attempts were made to estimate HNF6 protein levels in healthy or colonic tumor tissues. As denoted above, FOXA2 interacts with HNF6. Therefore, we investigated FOXA2 expression in the healthy colon and colonic tumors and occasionally observed in healthy or colonic cancer a faint immunoreactive band, an example of which is shown in Fig. 10D
. Recently the formation of a C/EBP-HNF6 protein complex was reported to stimulate HNF6 transcriptional activity (22)
. In our study, the C/EBP protein in healthy liver or colorectal liver metastases was below the level of detection (Fig. 10C
).
|
|
Electrophoretic mobility shift assays
We used an optimized oligonucleotide probe to investigate DNA binding of HNF6. We observed abundant DNA binding with nuclear extracts of healthy liver and confirmed the specificity for HNF6 by shifting the band with an HNF6 antibody. HNF6 DNA binding was studied in competition assays with unlabeled probes to further demonstrate selectivity (Fig. 11
A). Strikingly, when nuclear extracts of colorectal liver metastases were used, no binding of HNF6 was observed (Fig. 11B
). Likewise, when probed with nuclear extracts isolated from healthy colon, no HNF6 band was observed. Thus, DNA binding activity of HNF6 is confined to nuclear extracts of healthy liver tissue.
|
For its well-known regulation by HNF6, we studied DNA binding of the NGN3 protein using nuclear extracts of liver and colon. This protein is required for endocrine cell fate and development of intestinal and gastric epithelium. HNF6 is an upstream activator of neurogenin 3, and expression of NGN3 depends on HNF6 binding to promoter sites of NGN3. Mice lacking HNF6 display severely reduced NGN3 expression and a reduced number of endocrine cells (17)
. We found NGN3 DNA binding not to differ among healthy liver and/or colorectal liver metastases (Fig. 12
A).
|
Finally, we probed for isoform specificity of HNF6 DNA binding and used published probes to distinguish among the different HNF6 variants. We used oligonucleotide probes optimized for HNF6 binding using promoter sequences of transtyrethin (TTR), hepatic NF 3 ß (=FOXA2), hepatic NF 4 (HNF4), and phosphoenolpyruvate carboxykinase (PEPCK) to differentiate among HNF6 isoforms (23)
. Binding affinity of HNF6ß was reported to be greater for the FOXA2 probe, whereas binding affinity of HNF6 is greater in the case of the HNF4 and PEPCK probes (23)
. As shown in Fig. 13
, HNF6 binds to an optimized HNF6 probe (see lane 1). Once again, specificity of binding was confirmed by using an HNF6 antibody (see lane 2, supershifted band). Competition assays with unlabeled probes for TTR, FOXA2, HNF4, and PEPCK were performed to differentiate among relative binding affinities of the highly homologous HNF6 and ß isoforms. As can be seen from Fig. 13
, competition of the HNF6 band was achieved for all but the PEPCK probe. Overall, we were unable to determine differences in relative binding affinities for HNF6 isoforms. Unfortunately, no commercial antibody is available to distinguish between these highly homologous isoforms of HNF6.
|
We also studied binding of HNF6 to promoter sequences within the TTR, FOXA2, HNF4, and PEPCK genes (see Materials and Methods for the genetic algorithm applied to predict the exact location of the HNF6 sites) using nuclear extracts of healthy human liver. Notably, binding of HNF6 to the TTR probe was most abundant. In the case of TTR and FOXA2, binding of HNF6 was confirmed by use of an HNF6 antibody. No HNF6 band could be supershifted with the HNF4 and/or PEPCK probes. In the case of PEPCK, results agree well when findings from the competition assay (see lane 6) are compared with HNF6 DNA binding studies (lanes 13, 14), but results with the HNF4 probe are less obvious (see lanes 5, 11, 12).
We investigated FOXA2 DNA binding with nuclear extracts of healthy liver, colorectal liver metastases, healthy colon, and colonic tumor. Essentially, no difference in DNA binding was observed, albeit FOXA2 DNA binding was less abundant with nuclear extracts from healthy or cancerous colonic tissue (see Fig. 12B, C
).
Expression of HNF6 target genes
As detailed in Materials and Methods, 24 liver metastases as well as 10 healthy livers were used in our microarray study. We probed specifically for expression of HNF6-targeted genes. The selection of HNF6 gene targets is based on the study by Odom et al., who used chromatin immunoprecipitation followed by DNA/DNA hybridization (CHIP-chip assay) to determine HNF6 binding to promoter sequences of the human liver genome (16)
. Table 5
gives an account of the 28 genes targeted by HNF6. The selection is based on major biological functions and includes genes coding for general metabolic functions, protein synthesis, transport, receptor, apoptosis, and promoter of tumor growth. As shown in Table 5
, most of the genes targeted by HNF6 were highly repressed, albeit at different levels. It is of considerable importance that three genes involved in the promotion of tumor growth were significantly elevated: chemokine ligand 1, chemokine ligand 3, and fatty acid binding protein).
|
DNA mutation analysis of HNF6 CBP binding domain and acetylation sites
As described above, no DNA binding of HNF6 was observed with nuclear extracts of colorectal liver metastases. This prompted our interest in studying sequence variations in the DNA binding domains of HNF6. After PCR amplification, we studied the coding sequences of the cut and homeodomains and subjected amplification products to direct sequencing using capillary electrophoresis. No sequence variations were found in the DNA binding domains (cut and homeodomains) when genomic DNA of healthy human liver and colorectal liver metastases were compared (see Fig. 14
). We display electropherograms of n = 3 representative patients, showing sequences of genomic DNA extracts of healthy human liver and colorectal liver metastases of an HNF6 acetylation site within the cut domain. The codon AAA at position 1015–1017 codes for a lysine. When substituted for arginine (K339R mutant protein), CBP acetylation is abrogated (21)
. In human hepatoma cells, the mutant HNF6 protein fails to accumulate and is transcriptionally inactive. It is important that such mutant protein cannot be stabilized by inhibiting ubiquitin proteasomal degradation (21)
. Thus, abrogation of HNF6 DNA binding cannot be explained by mutations in the cut and/or homeodomains.
|
Hierarchical gene cluster analysis of liver-enriched transcription factors and some target genes
We applied a hierarchical cluster method to a group of genes on the basis of similarity of expression. The cluster diagram is shown in Fig. 15
and is based on 783 gene expression results. Despite its complex nature, the clustering analysis provided a remarkable order. With a single exception, gene expression segregated clearly between healthy and cancerous tissue. Furthermore, the applied algorithm clearly distinguished among patients who received surgery to the right or left hemicolon. This suggests regional differences in the expression of liver-enriched transcription factors of the colon and allowed us to trace descendent tumor cells back to the anatomical region of the primary tumor. The algorithm further distinguished among subgroups within our patient population as shown in Fig. 15
. Furthermore, the dentogram shown on the right-hand side of Fig. 15
indicates a complex cluster that groups distinct sets of genes, with the mitochondrial ATPase (housekeeping gene) being distinctly separated from all other genes.
|
We observed groups of sets of genes, including HNF4 and some of its splice variants, HNF1 and HNF4 (note there is coregulation between HNF1 and HNF4), HSP90 and HSP105, among others. Clearly, gene cluster analysis is useful in identifying potential networks of regulated genes (see below).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 25)
. Among our patients, the time between resection of primary tumor and colorectal liver metastases was 14 months and did not differ between genders. Initially, we investigated gene expression of major liver-enriched transcription factors, as summarized in Table 2
. Of all transcription factors investigated, HNF6, HNF1ß, and C/EBP were statistically significantly regulated in metastatic tumors of the liver. This was unexpected, as HNF6 transcript expression was below the limit of detection in healthy tissue or primary tumors of the colon. There was no obvious relationship between HNF6 gene expression and tumor staging, although a statistically significant Spearman’s correlation coefficient could be determined for HNF6 gene expression in healthy liver compared with colorectal liver metastases.
Here we provide evidence for migrating tumor cells grown in a hepatic environment to result in pleiotropic expression of HNF6. Transcript expression of HNF6 was less in colorectal liver metastases than with healthy liver tissue, even though the coded protein was abundantly expressed, as shown in Fig. 9
. The organ-specific nutrient and growth factor supply of the liver likely regulated the expression of the otherwise silent HNF6. Furthermore, our Western blot experiments revealed major differences in the post-translational processing of the HNF6 protein. We observed two distinct immunoreactive bands for HNF6 with nuclear extracts of healthy liver and colorectal liver metastases. The acetylated HNF6 accounted for the upper immunoreactive band whereas the lower band is the unacetylated form. Unlike healthy liver with colorectal liver metastases, unacetylated HNF6 was prominent. Retention of HNF6 acetylation, however, is required for HNF6 transcriptional activity (21)
.
This prompted us to examine the expression of genes targeted by HNF6. We studied transcript expression of 28 bona fide genes targets, which we found to be mainly repressed. Table 5
provides an account of the regulated genes. Because of its complexity, a discussion of the oncogenomics of the several hundred genes regulated in colorectal liver metastases will be the subject of a separate report. Moreover, there is clear evidence for HNF6 to function as a coactivator protein and to potentiate transcriptional activity of FOXA2-targeted genes. Because of their coactivator interaction, we investigated FOXA2 protein expression in healthy liver and colorectal liver metastases, and found expression of FOXA2 protein to be highly induced in nuclear extracts of colorectal liver metastases. FOXA2 has been shown to inhibit HNF6 DNA binding to result in transcriptional repression of genes targeted by HNF6 (26)
. Therefore, the repressed transcript expression of HNF6-targeted genes is likely to be the result of an inhibitory protein interaction of FOXA2-HNF6, as reported elsewhere (26)
. Furthermore, inhibition of the histone acetyltransferase activity abrogated HNF6 synergistic activation of FOXA2-dependent transcription. Such inhibition, however, did not influence HNF6 transcriptional activation of HNF6-dependent genes (21)
. In addition, HNF6 interacted with the FOXA2 protein when bound to DNA to facilitate recruitment of p300/CBP proteins for transcriptional synergy of FOXA2-dependent genes (21)
. Although the FOXA2-HNF6 transcriptional synergy requires acetylation by the CREB binding protein coactivator, the stability of HNF6 is controlled equally by CBP-dependent acetylation. We therefore investigated sequence variations in the lysine codon acetylation site and in the DNA binding domain of HNF6, but did not identify sequence mutations (see Fig. 14
). We thus rule out sequence alterations as a molecular mechanism and propose abrogation of HNF6 DNA binding to be the result of an inhibitory protein interaction with FOXA2. In the study by Lannoy et al. (23)
, regions outside the DNA binding domain (amino-terminal half of HNF6) did not contribute to DNA binding and could therefore be deleted without affecting DNA binding activity. In another study by the same group (27)
, further deletions in the amino-terminal half did not affect DNA binding. In our study, we observed full-sized HNF6 that was unable to bind to DNA, but did not contain mutations in the DNA binding domain or the lysine codon acetylation site.
Indeed, FOXA2 was strongly induced in colorectal liver metastases, as shown in Fig. 10A
. Furthermore, HNF6 serves as a coactivator protein to enhance FOXA2 transcription. We observed binding of HNF6 to promoter sequences of FOXA2 (see Fig. 13
, lanes 9 and 10) with nuclear extracts of healthy liver. FOXA2 binding to a TTR promoter site (–111 to –85) was also observed with nuclear extracts of healthy liver and colorectal liver metastases, as well as healthy colon and colonic cancer (see Fig. 12B, C
). Although FOXA2 protein expression was strongly induced in liver metastases, its expression was variable in healthy and tumoral colonic tissue (see Fig. 10A
, D).
Taken collectively, expression of HNF6 is exceptional for colorectal liver metastases. Neither healthy nor primary colonic tumor tissue expresses HNF6. We show unacetylated HNF6 to be strongly expressed in colorectal liver metastases, but the protein is unable to bind to HNF6 recognition sequences. Likely, abrogation of HNF6 DNA binding is due to an interaction with FOXA2, which we found to be highly induced in colorectal liver metastases.
In general, hepatocyte nuclear factors play an essential role in determining cellular differentiation. These regulatory proteins function in a networked environment and bind to recognition sequences of targeted genes, providing regulatory chains where one hepatic NF may activate another. Such regulatory loops are of particular importance in the onset and progression of disease. For instance, HNF6 binds to cognate recognition sequences of HNF4 in order to regulate its expression (28)
. Recent evidence suggests loss of HNF4 expression to be an important determinant of HCC progression (7)
. In the study by Lazarevich et al., expression of liver-enriched transcription factors differed in slow- and fast-growing hepatocellular carcinomas. We found most of the liver-enriched transcription factors to be expressed in this secondary malignancy of the liver, albeit at different levels. We show expression of HNF6 in colorectal liver metastases to be the consequence of the nutrient and growth factor supply of the liver. To the best of our knowledge, this is the first report to demonstrate a gene environment interaction in metastatic disease.
The relationship between HNF6 and FOXA2 is controversial. In the study by Rausa et al., HNF6 functions as a coactivator protein to potentiate the transcriptional activity of Foxa2 (26)
whereas in the study by Rubins et al., HNF6 function is largely independent of Foxa2 (29)
. We observed massive induction of FOXA2, an inhibitory protein for HNF6 DNA binding, in colorectal liver metastases. Our study is the first report on the importance of liver-enriched transcription factors in a secondary liver malignancy. Of all transcription factors investigated, HNF6 and FOXA2 appeared to play a pivotal role, with the HNF6 gene and protein expression being driven by the hepatic nutrient supply of the liver. We demonstrate HNF6 DNA binding to be selectively abrogated and observed full-sized but unacetylated HNF6. This may point to a molecular mechanism by which FOXA2 inhibited HNF6 DNA binding. Our findings may well translate to novel approaches in therapy. Restoring HNF6 DNA binding activity is a likely remedy and may be achieved by targeting the FOXA2 protein in order to prevent disease progression and growth of colorectal liver metastases.
|
ACKNOWLEDGMENTS |
|---|
Received for publication June 20, 2006. Accepted for publication December 25, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
