Overexpression of far upstream element binding proteins: A mechanism regulating proliferation and migration in liver cancer cells

Authors


  • Potential conflicts of interest: none.

  • The data in this study were presented in part at the 21st European Congress of Pathology (Istanbul, 2007), and the authors were awarded the George Tiniakos Award.

Abstract

Microtubule-dependent effects are partly regulated by factors that coordinate polymer dynamics such as the microtubule-destabilizing protein stathmin (oncoprotein 18). In cancer cells, increased microtubule turnover affects cell morphology and cellular processes that rely on microtubule dynamics such as mitosis and migration. However, the molecular mechanisms deregulating modifiers of microtubule activity in human hepatocarcinogenesis are poorly understood. Based on profiling data of human hepatocellular carcinoma (HCC), we identified far upstream element binding proteins (FBPs) as significantly coregulated with stathmin. Coordinated overexpression of two FBP family members (FBP-1 and FBP-2) in >70% of all analyzed human HCCs significantly correlated with poor patient survival. In vitro, FBP-1 predominantly induced tumor cell proliferation, while FBP-2 primarily supported migration in different HCC cell lines. Surprisingly, reduction of FBP-2 levels was associated with elevated FBP-1 expression, suggesting a regulatory interplay of FBP family members that functionally discriminate between cell division and mobility. Expression of FBP-1 correlated with stathmin expression in HCC tissues and inhibition of FBP-1 but not of FBP-2 drastically reduced stathmin at the transcript and protein levels. In contrast, further overexpression of FBP-1 did not affect stathmin bioavailability. Accordingly, analyzing nuclear and cytoplasmic areas of HCC cells revealed that reduced FBP-1 levels affected cell morphology and were associated with a less malignant phenotype. Conclusion: The coordinated activation of FBP-1 and FBP-2 represents a novel and frequent pro-tumorigenic mechanism promoting proliferation (tumor growth) and motility (dissemination) of human liver cancer cells. FBPs promote tumor-relevant functions by at least partly employing the microtubule-destabilizing factor stathmin and represent a new potential target structure for HCC treatment. (HEPATOLOGY 2009.)

Microtubules are cytoplasmic components of the cytoskeleton that play a critical role in the maintenance of cell morphology, division, intracellular transport, and motility. Microtubule polymers consist of α- and β-tubulin heterodimers and are characterized by continuous transition between phases of shrinkage (depolymerization/catastrophe) and elongation (polymerization/rescue). This dynamic instability is regulated by several factors, including microtubule-associated proteins and microtubule destabilizers such as stathmin.1 The cytosolic phosphoprotein stathmin (oncoprotein 18) is expressed in most proliferating cells, and its concentration increases during the S-phase of the cell cycle. In this context, stathmin facilitates microtubule shortening by tubulin sequestration and active promotion of microtubule catastrophe, depending on the sequential phosphorylation status of four serine residues.2 When cells enter mitosis, stathmin is inactivated by phosphorylation, allowing the formation of the mitotic spindle; subsequent reactivation of stathmin by dephosphorylation is necessary for the exit from mitosis.1

Elevated expression of stathmin has been reported in several human malignancies such as breast cancer and hepatocellular carcinoma (HCC),3, 4 suggesting that stathmin-dependent microtubule dynamics could support the final stages of tumor cell mitosis. In addition, functional analyses revealed promigratory effects of stathmin in sarcoma and HCC cells,4, 5 suggesting that stathmin overexpression promotes tumor cell dissemination and thus invasiveness as well as the metastatic behavior of cancer cells. Especially in HCCs, elevated stathmin expression has been linked to increased tumor size, the presence of vascular invasion, intrahepatic metastasis, and decreased patient survival.6 Microtubule destabilizing factors represent potential therapeutic target structures, because reduced but also increased stathmin bioavailability sensitizes tumor cells to treatment with chemotherapeutic agents.4, 7

So far, the regulation of stathmin has been attributed to the modulation of chromatin structure by enhancer of zeste homologue 2 (EZH2),8 posttranscriptional silencing by the small noncoding RNA miR-223,9 as well as the expression of the tumor suppressor gene p53.10 In addition, we have recently demonstrated that specific gain of function mutations in p53 lead to the constitutive activation of stathmin in HCC cells.4

In order to further characterize the cellular network that regulates stathmin and thus microtubule-mediated protumorigenic functions in hepatocarcinogenesis, we demonstrate that far upstream element (FUSE) binding protein (FBP) family members partly govern regulatory functions of stathmin affecting distinct tumor cell functions. The close interplay of different FBP family members might represent a new molecular decision making process discriminating between cell division and motility.

Abbreviations

cDNA, complementary DNA; DN, dysplastic nodule; EZH2, enhancer of zeste homologue 2; FBP, far upstream element binding protein; FUSE, far upstream element; HCC, hepatocellular carcinoma; HGF, hepatocyte growth factor; mRNA, messenger RNA; PCR, polymerase chain reaction; siRNA, small interfering RNA; TMA, tissue microarray.

Materials and Methods

Cell Culture, Small Interfering RNAs, Plasmids, Transfection, and Sequences.

Culture of HCC cell lines and transient transfection of small interfering RNAs (siRNAs) has been described (final concentration for FBP-1 and FBP-2 siRNAs: 20 nM).11 All siRNA transfections were performed using two different sequences for each gene. For transient overexpression of FBP-1, complementary DNA (cDNA) was amplified by way of polymerase chain reaction (PCR) and subcloned in the eukaryotic expression vector pcDNA3.1 with primers containing 5′ HindIII and 3′ EcoRV restriction sites. The construct was verified by sequencing. Cells were transfected with 1 μg of plasmid DNA using Fugene 6 according to the manufacturers' instructions (Roche Diagnostics, Grenzach, Germany). Antibodies and respective dilutions as well as sequences of the primers and siRNAs used in this study are listed in Supporting Table 1.

Sample Preparation, Real-time Polymerase Chain Reaction, and Western Blot Analyses.

Total RNA isolation of 27 fresh frozen HCC samples and healthy liver tissues for semiquantitative real-time PCR was performed using the NucleoSpin RNA II kit according to the manufacturers' protocol (Macherey-Nagel, Duren, Germany). Total protein extracts were isolated using the 10× Cell-Lysis-Buffer (Cell Signaling/New England Biolabs, Frankfurt, Germany). The isolation of protein extracts from primary human HCC samples (n = 27),12 cDNA synthesis, and real-time PCR have been described.4 For western blot analyses, total protein extracts were separated, electro-transferred, and detected as published.4 For normalization, protein amounts of tissue extractions (30 μg) were determined using the Bradford assay.

Cell Viability, Proliferation, Apoptosis, and Migration.

All functional assays were performed 3 (proliferation, apoptosis, migration) or 4 days (viability) after transient transfection of siRNAs. Cell viability was measured using the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium-bromide) assay. Ninety-six hours after seeding, the medium was removed and the MTT solution (0.5 mg/mL in medium) was added. After 2 hours, the MTT solution was removed and tetrazolium salt was resolved in 100 μL of DMSO/ethanol solution (1:2). Colorimetric measurement was performed at 570 nm using an ELISA reader. Cell proliferation was determined using a bromodeoxyuridine enzyme-linked (BrdU) immunosorbent assay (Cell proliferation ELISA Biotrak, GE Healthcare/Amersham, Freiburg, Germany) according to the manufacturers' instructions. Tumor cell apoptosis was analyzed by poly(ADP-ribose) polymerase (PARP) cleavage using western immunoblotting and by way of caspase-3 activity assay. The migratory ability of cells was analyzed as described.11

Tissue Microarray Analysis, Immunohistochemistry, and Immunofluorescence.

The tissue microarray (TMA) contained two representative areas (diameter: 0.6 mm) of 15 healthy liver samples, 35 dysplastic nodules (DNs), and 161 HCCs (grading: 34 × G1, 96 × G2, 31 × G3). The samples were surgical cases previously resected at the University of Heidelberg and histologically classified according to the established criteria by two experienced histopathologists (P. S. and S. S.). The study was approved by the institutional ethics committee (application no. 206/05) of the Medical Faculty at Heidelberg University. Furthermore, a second patient cohort was analyzed for FBP-1/2 expression. Therefore, a TMA was constructed containing specimens from 61 consecutive HCCs (grading: 7 × G1, 44 × G2, and 10 × G3) that were surgically resected at the University Hospital Zurich, Switzerland, and histologically classified by two experienced pathologists (A. W. and M.-O. R.). Follow-up data for all patients were available. The study was approved by the local ethics committee (Kantonale Ethikkommission Zürich, application no. 26–2005). Immunohistochemistry using appropriate antibodies (anti–FBP-1/2 and anti-stathmin) and evaluation of FBP-1/2 as well as stathmin staining intensity were performed as described.4, 13

For immunofluorescence staining, transfected cells were seeded on coverslips and washed with phosphate-buffered saline and fixed with ice-cold methanol and acetone for 10 and 5 minutes after 2 days. The cells were incubated with anti–α-tubulin antibody for 1 hour and subsequently labeled with a secondary Cy3-linked anti-mouse antibody. Coverslips were mounted with Vectashield mounting medium with DAPI (Vector Laboratories, Burlingame, CA). Representative cells were digitally documented using the Leica TCS SL fluorescence microscope system (Leica, Wetzlar, Germany). Antibodies and respective dilutions used in this study are listed in Supporting Table 1.

Statistical Analysis and Software.

Data are presented as the mean ± standard deviation. The Spearman rank coefficient was used as a statistical measure of association. The statistical comparison between two groups was accomplished with the nonparametric Mann-Whitney U test, and the significance levels were defined as *P < 0.05, **P < 0.01, and ***P < 0.001. Overall survival was measured from the time of surgery and analyzed using the Kaplan-Meier method. Differences between survival curves were evaluated using two-sided log-rank statistics (software: SPSS).

For the densitometric quantification of signals, the AlphaEase FC software was used (V4.0; Alpha Innotech/Biozym, Hess. Oldendorf, Germany). Area determination was performed using an analysis software imaging system (Olympus, Hamburg, Germany).

Results

FBP-1 and FBP-2 Are Frequently Overexpressed in HCC Cells and Correlate with Tumor Progression and Survival.

In order to identify essential modifiers that regulate stathmin overexpression in tumor cells, we reanalyzed genome-wide comparative transcript profiles of human HCC in comparison with normal liver samples.14 Including the entry criteria of nucleic acid-binding capacity and dysregulation in >50% of HCCs, a significant coregulation between the expression of stathmin and levels of the single-strand DNA binding factors FBP-1 (r = 0.547, P <0.001) and FBP-2 (r = 0.429, P = 0.008) was identified.

To examine the expression of FBP-1 and FBP-2 in the course of human hepatocarcinogenesis, HCC tissue microarrays were analyzed by immunohistochemistry. Four different antibodies tested in our study exhibited reactivity for FBP-1 as well as FBP-2, whereas no antibodies exclusively detected FBP-1 or FBP-2 (Fig. 1A). Whereas nonparenchymal cells were positive for FBP-1/2, no significant nuclear staining was detected in hepatocytes of normal liver tissue (Fig. 1B). In premalignant DNs, a gradual increase of the nuclear FBP-1/2 expression was observed, whereas carcinomas exhibited an intense FBP-1/2 expression. Further, 63.2% of DNs and 78.9% of all carcinomas were scored as highly positive for FBP-1/2. When normal livers, premalignant DNs, and HCCs were considered, elevated FBP-1/2 expression significantly correlated with the process of malignant transformation (r = 0.351, P < 0.0001) (Fig. 1C). Analyses of an independent HCC cohort containing clinical follow-up data not only confirmed FBP-1/2 overexpression in HCCs but also revealed that high FBP-1/2 expression levels significantly correlated with poor cumulative survival of HCC patients (P < 0.05) (Fig. 1D). No correlations were found with tumor etiology (e.g., chronic viral hepatitis) and other parameters (e.g., sex, age, tumor staging). In the group of HCCs, the accumulation of FBP-1/2 positively correlated with tumor cell proliferation as determined by Ki-67 staining (r = 0.542, P < 0.001).

Figure 1.

Elevated expression of FBP-1 and FBP-2 in HCC tissues. (A) siRNA-mediated knockdown of FBP-1 and FBP-2 in HCC cells demonstrating antibody specificity for both proteins. (B) Immunohistochemical analyses of FBP-1/2 on TMAs containing non-neoplastic liver tissues, premalignant lesions (DN), and HCC (grading: G1, G2, and G3). Arrows indicate FBP-1/2 staining in non-parenchymal cells (magnification ×200). (C) Boxplot summarizing expression of FBP-1/2 in normal liver, DNs, and HCCs (grading: G1, G2, and G3). Asterisks and open circles indicate the upper and lower outlier, respectively. (D) Correlation of FBP-1/2 expression in HCC tissues with patient survival.

In order to confirm the overexpression of FBP-1 and FBP-2 in HCC tissues, we performed gene-specific semiquantitative real-time PCR for FBP-1 and FBP-2 in an independent set of HCCs (n = 27) and normal livers (n = 3). These results confirmed the microarray as well as the TMA datasets and revealed an elevated expression of FBP-1 (81.5%, 22/27) and FBP-2 (67%, 18/27) in the majority of HCCs (Fig. 2A). No reduction of FBP-1 and FBP-2 at the transcript level was detected as compared with normal liver tissues. Surprisingly, the expression profiles of both FPBs significantly correlated with each other (r = 0.82; P < 0.001).

Figure 2.

Coordinated overexpression of FBP-1 and FBP-2 in HCC tissues. (A) Relative quantification of FBP-1 (white bars) and FBP-2 (black bars) transcript levels was performed in 27 primary human HCCs. Relative expression changes were calculated in comparison to a mixture of three normal liver tissues. Bars show relative expression levels of FBP-1 and FBP-2 of each HCC arranged according to descending FBP-1 expression values. (B) Exemplary demonstration of FBP-1 and FBP-2 protein expression in HCC tissues (n = 27), cell lysates of primary human hepatocytes, and normal liver tissue. Normalization of protein loading was performed using Bradford quantification.

In order to further substantiate FBP-1 and FBP-2 overexpression in HCCs, protein expression was determined in the same set of HCCs by way of western immunoblotting. These experiments revealed a strong detectable signal for FBP-1 in 74% (20/27) and FBP-2 in 70% (19/27) of all HCCs, while in cell lysates of normal liver tissue and cultured primary human hepatocytes, FBP-1 and FBP-2 concentrations were very low (Fig. 2B).

Increased FBP-1 and FBP-2 Expression Supports Proliferation and Migration of Tumor Cells.

When we compared protein levels of FBP-1 and FBP-2 in different types of cancer cell lines (including cell lines derived from HCC, lung cancer, and squamous cell carcinoma), in all tested samples the concentrations were higher than those seen in primary human hepatocytes (Supporting Fig. 1A). In order to determine whether increased FBP-1/2 expression was of functional relevance in hepatocarcinogenesis, FBP-1 and FBP-2 expression was knocked down using two independent siRNAs in different HCC cell lines (Fig. 3, Supporting Fig. 2). For both target genes, a significant reduction was demonstrable at the transcript level (data not shown) and protein level (Fig. 3A). Surprisingly, the efficient inhibition of FBP-2 led to significantly elevated transcript (data not shown) and protein levels of FBP-1, whereas no effects of FBP-1–specific siRNAs on FBP-2 expression were observed. This regulatory connection was confirmed using four different FBP-2–specific siRNAs in several tumor cell lines (HCC and NSCLC cells, data not shown).

Figure 3.

Inhibition of FBP-1 and FBP-2 expression through RNA interference reduces tumor cell proliferation or motility. (A) Inhibitory efficiency of gene-specific siRNAs for FBP-1 and FBP-2 after transient transfection of Hep3B cells by western blot analyses compared with respective controls (untreated and nonsense siRNA-transfected cells). Expression of each FBP was separately knocked down by two independent siRNAs. (B) Tumor cell viability after FBP-1 (left columns) and FBP-2 (right columns) knockdown was measured 4 days after seeding. (C) Tumor cell proliferation was determined based on bromodeoxyuridine incorporation 3 days after FBP-1 (left) and FBP-2 (right) inhibition. (D) HCC cell motility was determined using a two-dimensional migration assay 3 days after FBP-1 (left) and FBP-2 (right) knockdown. Prior to induction of migration by HGF, tumor cell proliferation was blocked using mitomycin-c. For all functional assays, untreated cells were used for calibration and nonsense siRNA-transfected cells were used for statistical evaluation.

The transient knockdown of FBP-1 was associated with decreased tumor cell viability (Hep3B: up to −47%) (Fig. 3B). These biological consequences after FBP-1 inhibition correlated with diminished tumor cell proliferation (Hep3B: up to −59%) (Fig. 3C), while no significant effects on apoptosis were detected (Supporting Fig. 1B,C). Only moderate effects were observed concerning tumor cell viability and proliferation after inhibition of FBP-2. Equally, no consequences on cell death were seen after the reduction of FBP-2 concentrations.

The basic motility under nonstimulated conditions of both analyzed tumor cell lines is generally low (e.g., HuH-7: <10% motile cells with 6 μm/hour).11 In order to more closely reflect the natural situation, we stimulated the tumor cells with promotogenic hepatocyte growth factor (HGF). Paracrine activation of the HGF/MET signaling pathway itself, which is well-documented in human HCC,15 did not affect FBP-1 or FBP-2 expression in HCC cells (data not shown). The effects of FBP-1 and FBP-2 knockdown on HGF-stimulated HCC cell motility were analyzed after reduction of cell division by mitomycin-c in two-dimensional scratch assays (Fig. 3D). After wounding of the tumor cell layer, motility-based recovery of the scratched area was not significantly affected in samples with diminished FBP-1 expression, but drastic effects were detectable after FBP-2 inhibition. All functional results were reproduced with another HCC cell line (HuH-7, Supporting Fig. 2).

Elevated FBP-1 Induces Stathmin Expression in HCC Cells.

Based on the strong correlation between FBP-1 and stathmin detected in the HCC profiling experiments as well as the significant effects after FBP-1 inhibition on tumor cell proliferation, we analyzed whether FBP-1 is a regulator of stathmin expression in HCC cells.

To test whether FBP-1 and stathmin are coexpressed in HCCs, correlative analyses with TMAs were performed, showing that nuclear accumulation of FBP-1/2 correlated with increased cytoplasmic staining for stathmin (r = 0.405; P <0.001) (Fig. 4A). To more precisely analyze FBP-1 and stathmin expression in HCC tissues, we performed semiquantitative real-time PCR and observed that elevated FBP-1 mRNA levels significantly correlated with stathmin transcript levels (r = 0.547; P <0.001) (Fig. 4B). Equally, FBP-2 mRNA levels correlated with stathmin transcript concentrations (r = 0.429; P <0.001) (data not shown).

Figure 4.

FBP-1 and stathmin/oncoprotein 18 overexpression correlates in HCC cells. (A) Immunohistochemical analyses of stathmin and FBP-1/2 expression using HCC TMAs. Representative stainings show strong nuclear accumulation of FBP-1/2 and cytoplasmic staining for stathmin. No stathmin expression was detected in normal liver tissue (original magnification ×200). Bars, 100 μm. (B) Relative quantification of FBP-1 (black bars) and stathmin (grey bars) transcript levels was performed in 27 human HCCs. Relative expression changes were calculated in comparison to a mixture of three normal liver tissues. Bars were arranged according to descending stathmin expression values. (C) Vector-based overexpression of FBP-1 in Hep3B cells for 4 days did not affect stathmin levels. Expression intensity of FBP-1 and stathmin was normalized to respective actin levels. (D) siRNA-mediated knockdown of FBP-2 and subsequent accumulation of FBP-1 did not induce stathmin concentrations in Hep3B cells.

In order to determine the potential functional relevance of FBP-1 for stathmin expression, FBP-1 was transiently overexpressed in HCC cells (FBP1-pcDNA3.1). Although significant amounts of FBP-1 were detected 48-96 hours after transfection, no significant regulation of stathmin bioavailability was observed compared with appropriate controls (Fig. 4C). In addition, siRNA-mediated inhibition of FBP-2 associated with increased FBP-1 concentrations did not affect stathmin at the protein levels (Fig. 4D). All results were reproduced with another HCC cell line (HuH-7) (Supporting Fig. 3A,B). Together, these results demonstrate that elevated amounts of FBP-1 did not further induce stathmin in HCC cells.

In contrast, a clear inhibition of stathmin was achieved at the transcript and protein levels after transient transfection of FBP-1–specific siRNAs (Fig. 5A,B). Moreover, time-resolved analyses of FBP-1 and stathmin expression revealed that inhibitory effects at the FBP-1 protein level were detectable 36 hours after siRNA transfection, whereas diminished stathmin levels were observed after 42 hours, demonstrating stathmin as a downstream target of increased FBP-1 expression (Fig. 5C). Transient transfection of FBP-1–specific siRNA induced an effective and long-lasting reduction of stathmin for more than 6 days (data not shown). Results were reproduced with HuH-7 cells (Supporting Fig. 3C,D).

Figure 5.

FBP-1 regulates stathmin/oncoprotein 18 in HCC cells. (A) Relative quantification of FBP-1 (black bars) and stathmin (grey bars) transcript levels after FBP-1 knockdown. (B) FBP-1 inhibition reduces stathmin concentration at the protein level after transient transfection. (C) Kinetics of stathmin expression after FBP-1 knockdown. Thirty-six hours after transient transfection of FBP-1–specific siRNA, protein levels rapidly declined for up to 66 hours. The first detectable effects on stathmin expression were observed 42 hours after transfection. (D) Diminished HCC cell proliferation induced by FBP-1 knockdown, FCS withdrawal, and mitomycin-c treatment and effects on stathmin expression (left columns, proliferation index; right figure, stathmin protein levels).

In order to exclude that diminished stathmin levels represented an epiphenomenon due to FBP-1–mediated reduction of cell proliferation, mitosis was inhibited by FCS deprivation and mitomycin-c administration in Hep3B cells (Fig. 5D). As expected, FBP-1 inhibition, FCS withdrawal, and treatment with DNA-intercalating drugs significantly reduced proliferation in HCC cells. However, only FBP-1 knockdown reduced stathmin expression, demonstrating direct regulatory effects of FBP-1 on stathmin and independence of unspecific mechanisms coupled to mitosis. Data were reproduced with an additional cell line (HuH-7, Supporting Fig. 3E).

Based on the marked consequences of FBP-1 on stathmin expression, we hypothesized that the knockdown of FBP-1 next to proliferation also affects microtubule-dependent cell polarity and morphology. For this reason, the two-dimensional extension of HCC cell nuclei and cytoplasmic areas with and without FBP-1 inhibition were measured after α-tubulin immunofluorescence staining. Indeed, the cytoplasmic areas were significantly elevated after FBP-1 reduction (up to five-fold) with clear effects on cell morphology and an increased nuclear-to-cytoplasmic ratio (Fig. 6). These findings suggest a less malignant phenotype after the specific knockdown of FBP-1 and subsequent reduction of stathmin. Equal results were detected in HuH-7 cells (Supporting Fig. 4). Moreover, the number of polyploid cells was increased after FBP-1 inhibition in both cell lines compared with respective controls (Hep3B, increase from 3% to 27%; HuH-7, increase from 14% to 20%).

Figure 6.

FBP-1 modulates HCC cell morphology. Nuclear and cytoplasmic areas of tumor cells were measured after siRNA-mediated inhibition of FBP-1 in HCC cells. The relative ratio of total cell area to nucleus area is depicted. Representative cells are shown after immunofluorescence staining of α-tubulin. Untreated cells were used for calibration, and nonsense siRNA-transfected cells were used for statistical evaluation. Bar, 30 μm.

Lastly, it was analyzed whether FBP-1 affects stathmin concentrations through modulating previously described regulators, namely EZH2 and miR-223.8, 9 No significant reduction of EZH2 expression or induction of miR-223 were detected after efficient inhibition of FBP-1 in different HCC cell lines, demonstrating independence of these mechanisms (Supporting Fig. 5A-D).

Discussion

We have identified a regulatory mechanism for the protumorigenic microtubule-destabilizing factor stathmin. FBP-1 and FBP-2 were frequently elevated and significantly coexpressed with stathmin in human hepatocarcinogenesis. These data demonstrate that biological effects of coexpressed FBP-1 and FBP-2 in carcinogenesis are constituents of an oncogenic mechanism with prognostic impact.

FBP family members (FBP-1, FBP-2, and FBP-3) are multifunctional factors and have been linked with various transcript processing steps such as splicing,16 mRNA stabilization,17 and degradation.18 Because FBPs also recognize single-stranded DNA cis elements of genes and directly affect transcription, these factors are also regarded as transcription factors.19 In contrast to conventional transcription factors, FBPs bind torsionally stressed genomic regions of transcriptionally active genes after melting of so-called FUSE,20 which is a structurally unstable constituent of gene promoters. For this reason, the FUSE/FBP-system represents a direct (mechano-) sensor of gene activity. In this context, sequential binding of different FBP family members under physiological conditions may fine-tune the temporary expression of genes with low transcriptional activity or calibrate the expression kinetics of important key regulators such as the proto-oncogene c-myc.21, 22

The microtubule-destabilizing protein stathmin is indeed expressed only in small quantities by resting hepatocytes but is highly expressed by mitotic hepatocytes during regeneration after hepatic ischemia/reperfusion and partial hepatectomy and in hepatocarcinogenesis.4, 23 Therefore, stathmin fulfills all the requirements of a protein whose expression is temporarily induced in hepatocytes of normal liver tissue after stimulation of cell growth but is reduced to a basal level after cessation of stimuli. Although a fast effect on stathmin expression was observed after FBP-1 inhibition (Δ6 hours), no previously described FUSE site was found upstream of the first STMN1 exon (up to −2,000 bp).24 However, it has recently been demonstrated that the c-myc FUSE significantly varies from the perfect consensus sequence recognized by FBPs, pointing to the existence of additional sites that bind FBPs more strongly.25 Several segments within the STMN1 promoter indeed are closely related to the proposed optimal FBP binding sequence. The functional relevance of these segments in the FBP-1–driven regulation of stathmin must be confirmed experimentally.

Moreover, we demonstrated that activation of EZH2 expression or reduction of inhibitory miRNAs are not the relevant molecular mechanisms for FBP-1–dependent stathmin overexpression in HCC cells. Despite the fact that FBP-1 regulated stathmin mRNA quantity, the molecular mechanisms by which FBP-1 induced stathmin expression in HCC cells—especially direct transcriptional regulation versus regulation of transcript stability or degradation—is currently unknown.17, 18 FBP-1 knockdown affected microtubule-dependent effects, including tumor cell proliferation and maintenance of morphology, suggesting that reduced FBP-1–dependent expression of stathmin is functionally and morphologically associated with a less malignant phenotype. Surprisingly, only the efficient knockdown but not the overexpression of FBP-1 affected stathmin concentrations in different cell lines, indicating saturated FBP-1 levels in HCC cells with regard to the regulation of stathmin. More importantly, only FBP-1 inhibition—but not the FBP-2 knockdown—reduced stathmin concentrations, pointing to the existence of FBP family member–specific effector mechanisms in HCC cells. Differences in the bioactivity of FBPs are further supported by the fact that these family members significantly vary in intracellular velocity, transactivator strength, and binding capacity to p38/JTV-1, which targets FBPs for degradation.21, 26

FBP overexpression was recently detected in six HCCs using 2D-DIGE analyses, and it has been speculated that elevated FBP-1 concentrations, in addition to chromosomal gains of the c-myc locus (8q24), might further support c-MYC expression in HCC cells.27 However, we did not detect significant correlations between nuclear FBP-1 or FBP-2 and c-MYC expression in a large cohort of HCC tissues. Moreover, c-MYC transcript and protein levels were only moderately affected after inhibition of FBP-1 or FBP-2 in different HCC cell lines (data not shown), suggesting that FBP-dependent regulation of c-MYC plays a minor role in hepatocarcinogenesis. This is supported by previous studies demonstrating that HepG2 cells express c-MYC in a FUSE-independent manner.20 In addition, c-MYC but not stathmin has been identified as a target gene for FBP-1 in a cervix carcinoma cell line, again pointing to cell type–specific effects of FBP-1 on target gene expression.21

Inhibition of FBP-1 and FBP-2 predominantly affects different cell functions (proliferation and migration, respectively), suggesting protumorigenic effector mechanisms depending on the regulation of distinct target genes. This relationship is even more complex, because FBP-2 expression negatively regulates FBP-1 bioavailability in all analyzed cell lines. Because FBP-1 and FBP-2 are coexpressed, the combination of common upstream mechanisms as well as the regulatory interplay between FBPs contributes to specific FBP-1/2 concentrations in HCC cells. Therefore, the relative and the absolute intranuclear stoichiometries of both factors might represent a molecular decision-making process discriminating between cell division and cell movement.21 In this case, an excess of FBP-2 reduces FBP-1 levels and therefore switches tumor cell behavior from proliferation to migration. This biological concept of decision-making processes is well established, because cell division and cell movement are mutually exclusive processes for cells under defined (micro-)environmental conditions. For example, the concentration-dependent stimulation by growth factors such as the platelet-derived growth factor has been demonstrated to define the switch from a migratory to a proliferative phenotype,28 indicating that expression levels or stoichiometric relations of proteins discriminate between cell function. Therefore, depending on the respective exogenous stimuli, the ratio of FBP-1 and FBP-2 may in part influence the functional dichotomy responsible for tumor growth and metastasis.

Our data demonstrate that high-level expression of different FBP family members in hepatocarcinogenesis correlates with poor cumulative survival of HCC patients and promotes tumor growth and migration. Therefore, the coordinated activation of FBP-1 and FBP-2 in most HCCs (>70%, TMA data) represents a new target for pharmacological intervention. These possible therapeutic approaches range from inhibition of common positive upstream modulators for FBP-1/2 (e.g., growth factor receptors) to the activation of proteolytic degradation of FBPs. Indeed, p38/JTV-1 has been described to stimulate polyubiquitination and subsequent proteasomal degradation of FBP-1 in murine alveolar cells, also suggesting that aberrant proteolysis is one possible mechanism for FBP dysregulation in tumor cells.26 In addition, small chemical compounds may inhibit FBP/nucleic acid interaction and might therefore reduce FBP-mediated effects on transcription or mRNA stability. First approaches have been initiated to identify lead compounds that specifically recognize the hydrophobic pocket of FBP, which is essential for nucleic acid binding.29 Lastly, targeting bioactivity of functionally relevant downstream effectors (e.g., stathmin for FBP-1) might reduce FBP-driven effects in tumor cells. In this context, it was demonstrated that chemical modification of stathmin by nitrosoureas inhibits microtubule dynamics and reduces stathmin-dependent protumorigenic effects.30

Acknowledgements

We thank Sarah Messnard and Silvia Behnke for excellent technical assistance. Normal stomach, kidney, and colon tissues were obtained from the tissue bank of the National Center of Tumor Diseases, Heidelberg, Germany.

Ancillary