Potential conflict of interest: Nothing to report.
Lipocalin-2 negatively modulates the epithelial-to-mesenchymal transition in hepatocellular carcinoma through the epidermal growth factor (TGF-beta1)/Lcn2/Twist1 pathway
Article first published online: 7 AUG 2013
Copyright © 2013 by the American Association for the Study of Liver Diseases
Volume 58, Issue 4, pages 1349–1361, October 2013
How to Cite
Wang, Y.-P., Yu, G.-R., Lee, M.-J., Lee, S.-Y., Chu, I.-S., Leem, S.-H. and Kim, D.-G. (2013), Lipocalin-2 negatively modulates the epithelial-to-mesenchymal transition in hepatocellular carcinoma through the epidermal growth factor (TGF-beta1)/Lcn2/Twist1 pathway. Hepatology, 58: 1349–1361. doi: 10.1002/hep.26467
- Issue published online: 1 OCT 2013
- Article first published online: 7 AUG 2013
- Accepted manuscript online: 20 MAY 2013 04:30PM EST
- Manuscript Accepted: 12 APR 2013
- Manuscript Received: 20 DEC 2012
- National Research Foundation of Korea by the Ministry of Education, Science and Technology. Grant Number: 2011-0009814
- National R&D Program for Cancer Control. Grant Number: 0620220
- Korean Health Technology R&D Project. Grant Number: A101834
- Ministry for Health, Welfare and Family Affairs, Republic of Korea
- Top of page
- Materials and Methods
- Supporting Information
Lipocalin-2 (Lcn2) is preferentially expressed in hepatocellular carcinoma (HCC). However, the functional role of Lcn2 in HCC progression is still poorly understood, particularly with respect to its involvement in invasion and metastasis. The purpose of this study was to investigate whether Lcn2 is associated with the epithelial-mesenchymal transition (EMT) in HCC and to elucidate the underlying signaling pathway(s). Lcn2 was preferentially expressed in well-differentiated HCC versus liver cirrhosis tissues, and its expression was positively correlated with the stage of HCC. The characteristics of EMT were reversed by adenoviral transduction of Lcn2 into SH-J1 cells, including the down-regulation of N-cadherin, vimentin, alpha-smooth muscle actin, and fibronectin, and the concomitant up-regulation of CK8, CK18, and desmoplakin I/II. Knockdown of Lcn2 by short hairpin RNA (shRNA) in HKK-2 cells expressing high levels of Lcn2 was associated with EMT. Epidermal growth factor (EGF) or transforming growth factor beta1 (TGF-β1) treatment resulted in down-regulation of Lcn2, accompanied by an increase in Twist1 expression and EMT in HCC cells. Stable Lcn2 expression in SH-J1 cells reduced Twist1 expression, inhibited cell proliferation and invasion in vitro, and suppressed tumor growth and metastasis in a mouse model. Furthermore, EGF or TGF-β1 treatment barely changed EMT marker expression in SH-J1 cells ectopically expressing Lcn2. Ectopic expression of Twist1 induced EMT marker expression even in cells expressing Lcn2, indicating that Lcn2 functions downstream of growth factors and upstream of Twist1. Conclusion: Together, our findings indicate that Lcn2 can negatively modulate the EMT in HCC cells through an EGF (or TGF-β1)/Lcn2/Twist1 pathway. Thus, Lcn2 may be a candidate metastasis suppressor and a potential therapeutic target in HCC. (Hepatology 2013;58:1349–1361)
alpha-smooth muscle actin
epidermal growth factor
epidermal growth factor receptor
tetrazolium salt 3-(4,5-dimethylthiazol-2-yl)−2,5-diphenyl tetrazoliumbromide
Lipocalin-2 (Lcn2), also known as NGAL, belongs to the lipocalin protein family and was first purified from human neutrophils because of its association with gelatinase. Lcn2 can exist as a 25-kDa monomer, 46-kDa disulfide-linked homodimer, and/or 135-kDa disulfide-linked heterodimer with neutrophil gelatinase. Elevated Lcn2 expression has been observed in multiple human cancers including breast, colorectal, and ovarian cancers; however, the biological roles of elevated Lcn2 in cancer cells are not yet clear.[3-5] Substantial data indicate that Lcn2 is involved in invasion and metastasis. Lcn2 is able to facilitate gastrointestinal mucosal regeneration by promoting cell migration. In breast cancer, Lcn2 expression is considered to be a poor prognostic marker and is associated with tumor cell invasiveness. Its overexpression has been shown to increase cell migration, invasion, and lung metastasis in 4T1 murine breast cancer cells.[7, 8] However, other studies reported that Lcn2 suppressed cellular invasion and metastases in colon cancer and in Ras-transformed mouse mammary cells in vitro.[9, 10] Recently, Lcn2 was also shown to suppress invasion and angiogenesis in pancreatic cancer. Consistent with results from these previous studies, Lcn2 expression in ovarian cancer blocked the epithelial-to-mesenchymal transition (EMT), one of the hallmarks of invasive neoplasia.
In hepatocellular carcinoma (HCC), high expression of Lcn2 has been identified by microarray analyses, suggesting a potential role for Lcn2 in the development and progression of HCC.[12, 13] We previously reported that Lcn2 might prevent the progression of HCC by suppressing the proliferation and invasion of HCC cells by way of suppression of the JNK and PI3K/Akt signaling pathways. However, to date little is known about the role of Lcn2 in invasion and metastases during HCC progression. Increasing evidence indicates that aberrant activation of the embryonic programmed EMT plays a key role in tumor cell invasion and metastasis during tumor progression. EMT is a characteristic of most aggressive metastatic cancer cells.[15, 16] Cells that undergo EMT morphogenesis undergo a switch in phenotype from an apical-basolateral, polarized epithelial phenotype to a spindle-shaped, fibroblast-like, mesenchymal phenotype. A key feature in the initiation and execution of the EMT is down-regulation of E-cadherin (E-cad) expression. It was recently reported that EMT is promoted by interactions between the transcription factor Twist1 and epidermal growth factor (EGF) pathway in epidermal growth factor receptor (EGFR)-mutated lung adenocarcinoma.
In the present study we investigated the function of Lcn2 in HCC cell proliferation and invasion in vitro, and evaluated the role of Lcn2 in tumorigenicity and metastases in a mouse model system. We discovered that there is a correlation between Lcn2 expression and the loss of EMT characteristics in HCC cells, and found that Lcn2 can negatively modulate EMT in HCC cells through the EGF (or transforming growth factor beta1 [TGF-β1])/Lcn2/Twist1 pathway.
Materials and Methods
- Top of page
- Materials and Methods
- Supporting Information
Cell lines, including THLE-2, HepG2, Hep3B, Alexander (PLC/PRF/5), and SK-HEP-1, were obtained from the American Tissue Culture Collection (ATCC, Rockville, MD). Huh7 and Focus cells were acquired from the Korean Cell Line Bank (KCLB, Seoul, South Korea) and the MD Anderson Cancer Center (Dr. J-S. Lee, Houston, TX), respectively. SH-J1 (EMT phenotype), HLK-2 (epithelial phenotype), HKK-2 (epithelial phenotype), HLK-5 (epithelial phenotype), Choi-CK (epithelial phenotype), Cho-CK (epithelial phenotype), JCK (epithelial phenotype), and SCK cells (EMT phenotype) were established from HCC and cholangiocarcinoma tissues, cultured in Dulbecco's modified Eagle's medium (DMEM) medium (Sigma, St. Louis, MO) supplemented with 10% fetal bovine serum (FBS) (Invitrogen) and 1% L-glutamine, and grown at 37°C in the presence of 5% CO2, as described.[19-21] Tumor cell lines routinely used in HCC experiments are generally late passage cells that have been propagated numerous times, and are therefore prone to phenotypic changes and reduced expression of Lcn2. We therefore used early passage tumor cells with an epithelial phenotype, i.e., recently established tumor cell lines. The cell lines used in the present study are described in detail in Supporting Table S1.
Array Data Analysis
After the beadchips were scanned with the Illumina BeadArray reader, the microarray data were normalized using the quantile normalization method in the linear models for microarray data (LIMMA) package in the R language environment. Measured gene expression values were log2-transformed and the median was centered across genes and samples. We identified genes that were differentially expressed among the two classes using a random-variance t test. We next sought to identify a limited number of genes whose expression was tightly associated with the two subgroups. By applying a stringent threshold cutoff (P < 0.05 and 1.5-fold difference), we identified 2,446 features in the nontumor (NT)-HCC group and 1,399 features in the LC and GI/II groups. Cluster analysis was performed by calculating Pearson correlation coefficients and performing average linkage hierarchical clustering using Cluster and TreeView software.
Cell Migration Assay
Cell migration was measured using a cell wound-healing assay in six-well plates in culture medium containing DMEM with 10% FBS. Cells were grown to 90% confluence, rinsed with phosphate-buffered saline (PBS), and then starved for 24 hours in serum-free medium. A sterile 200 μL pipette tip was used to create three separate, parallel wounds, and migration of the cells across the wound line was assessed after 36 or 48 hours. Three independent experiments were performed.
In Vitro Cell Invasion Assay
Cellular invasiveness was quantified using a modified Matrigel Boyden chamber assay, as described.
Mouse Tumorigenicity Assay
SH-J1 cells were stably transfected with Lcn2 expression plasmid (Lcn2-7 or Lcn2-23), and 5 × 106 cells in 100 μL PBS were then inoculated subcutaneously into both shoulders of nude mice. Growth curves were plotted based on mean tumor volume within each experimental group at the indicated timepoints. Tumor length and width were monitored. Tumor volume was calculated according to the following equation: V (mm3) = width (mm2) × length (mm)/2. Tumor growth was observed for at least 8 weeks. In vivo tumorigenic experiments were performed using seven mice per treatment group.
In Vivo Metastasis Assay
A tail vein injection assay was used to assess the effect of Lcn2 on tumor metastasis. SH-J1-luc cells (5 × 105 cells in 200 μL PBS per mouse) previously transfected with either recombinant vector containing full-length Lcn2 or empty vector were injected into the tail veins of 6-week-old athymic nude mice. Mice were assessed for long-distance lung metastasis at 6 weeks (all seven mice per group). The number of lung metastasis nodules was counted to analyze the effects of treatment on spontaneous tumor metastasis.
Bioluminescence Imaging and Analysis
We established an orthotopic nude mouse lung metastasis model using SH-J1-luc cells stably expressing Lcn2. Briefly, 5 × 105 cells in 200 μL of PBS were injected into the tail veins of Balb/c nude mice (6 weeks old, n = 6). Tumor growth was monitored twice after 5 and 6 weeks by the Xenogen IVIS imaging system 100 (Caliper Lifescience, Hopkinton, MA; exposure time 10 s, level B/FOV15). Mice were anesthetized with 3% isoflurane after administration of 150 mg/kg body weight firefly D-luciferin (Caliper Lifescience) by intraperitoneal injection for imaging.
Quantification and Statistical Analyses
Bands on western blots and from semiquantitative reverse-transcription polymerase chain reaction (RT-PCR) were scanned using the LAS3000 system (Fuji Photo Film, Tokyo, Japan), and densitometric data were obtained by normalizing the results to the levels of actin and 18S. Statistical analyses were carried out using SPSS v. 16.0. Correlations between Lcn2 expression and clinicopathologic parameters or EMT marker expression was evaluated by the Kruskal-Wallis test or Spearman rank correlation test. The chi-square test and Fisher's exact test were used to compare variables between groups. Survival rates were calculated by the Kaplan-Meier method, and differences in survival curves were analyzed using the log-rank test. Data are expressed as averages ± standard deviation (SD). Differences were analyzed by dependent or independent t tests and P values ≤ 0.05 were considered significant.
- Top of page
- Materials and Methods
- Supporting Information
Expression of Lcn2 in HCC Tissues and Cell Lines
Expression analysis using gene expression data obtained from the publicly available Gene Expression Omnibus database (GEO1898 and GSE4024) revealed that Lcn2 expression in HCC tissue was significantly higher than that in matched nontumor surrounding liver tissues. To determine whether the microarray data derived from our Korean cohort was consistent with previous gene expression data, we performed BeadChip DNA microarray analysis of 42 HCCs and corresponding nontumor tissues. Unsupervised hierarchical clustering analysis of all tissues was based on similarities in the expression patterns for all genes (Fig. 1A). All tissue samples, except for five samples, clustered into one of two main groups: a nontumor liver tissue group (NT; normal liver + liver cirrhosis) or a tumor tissue group (HCC; Edmondson grades I-IV). In the two major sample clusters, genes with a P value <0.05 and with a mean difference of expression >1.5 between the two groups were selected. Lcn2 (2.05-fold) was preferentially expressed in HCC tissues compared with nontumor tissues. Next, using the same clustering analysis of four subgroups (NL, normal liver; LC, liver cirrhosis; well-differentiated HCC [Edmondson grade I/II]; poorly differentiated HCC [Edmondson grade III/IV]), we identified 380 genes out of 1,399 genes that were up-regulated in the well-differentiated HCC samples (including Lcn2; 2.2-fold). In contrast, 1,019 genes were down-regulated compared to LC (Supporting Table 4). The microarray data were registered with the Gene Expression Omnibus (GEO) database (Accession No. GSE36411). Additionally, to evaluate associations between Lcn2 expression and signaling pathways associated with the EMT, we performed network analysis using Ingenuity Pathway Analysis (IPA) software to generate an interaction network containing relevant biological information for the 1,399 genes (Supporting Fig. S1A). We identified 25 networks enriched for cell death and survival, cellular assembly and organization, cellular compromise, and hepatic system disease. Furthermore, supervised class comparison of the differentially expressed genes revealed that the top six significant canonical pathways (P < 0.05) were cancer, gastrointestinal disease, amino acid metabolism, small molecule biochemistry, tissue morphology, and cellular movement (Supporting Fig. S1B).
We also assessed Lcn2 mRNA expression by semiquantitative RT-PCR; 25 of 40 (62.5%) HCC specimens expressed higher levels of Lcn2 than adjacent nontumor liver tissue samples (Supporting Fig. S2). We further statistically analyzed Lcn2 messenger RNA (mRNA) abundance by real-time RT-PCR in six relevant groups of samples: normal liver (NL), liver cirrhosis (LC), GI, GII, GIII, and GIV HCCs (Fig. 1B). Levels of Lcn2 mRNA significantly increased according to the differentiation status of HCCs. However, Lcn2 expression in EMT-relevant GIV HCC was lower than that in GIII HCC, suggested that Lcn2 expression is significantly correlated with a worse differentiation grade, but negatively correlated with EMT in HCC. To determine whether Lcn2 expression is inversely correlated with the expression of EMT markers and regulators (EGF and TGF-β1), we measured mRNA levels of these markers by real-time RT-PCR (Supporting Fig. S3). Lcn2 expression was positively correlated with the expression of epithelial markers (DesI/II and CK8) and inversely correlated with the expression of mesenchymal markers (VIM and FN). However, expressions of endogenous EGF and TGF-β1 were positively correlated with Lcn2 expression. We also statistically evaluated if there were correlations between Lcn2 expression in HCC and clinicopathological variables (Supporting Table S5). Intriguingly, stage (AJCC) and Edmondson differentiation grade were the variables that showed significant differences among subgroups. Furthermore, survival analysis revealed that elevated expression of Lcn2 mRNA was significantly associated with a poor prognosis of short survival (Supporting Fig. S4).
Polyclonal rabbit antibody for Lcn2 was tested for specific immunoreactivity by transfecting HEK293T cells with GFP- or c-Myc-tagged expression plasmids (Supporting Fig. S5). Lcn2 protein expression was determined by immunoblot analysis in 12 HCC/normal tissue pairs and eight colon cancer/normal tissue sample pairs. Lcn2 was highly expressed in HCCs (58%) and colon cancers (100%) compared with nontumor tissues (Fig. 1C; Supporting Fig. S6). Expression of Lcn2 was also determined in a panel of HCC and cholangiocarcinoma (CC) cell lines (Fig. 1D). Lcn2 was highly expressed in HLK-2, HKK-2, and HLK-5 cells, but weakly expressed in THLE-2 immortalized hepatocytes. Intriguingly, HCC (SH-J1) and CC (SCK) cells that had undergone EMT barely expressed Lcn2. Intracellular Lcn2 existed as a 25-kDa protein, corresponding to monomeric Lcn2, as determined by reducing and nonreducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). In contrast, secreted Lcn2 existed in multiple forms, with 25, 40, 75, and 115/135 kDa bands corresponding to monomers, homodimers, homotrimers, and heterodimeric Lcn2/MMP-9 complexes detected on nonreducing gels (Supporting Fig. S7). Our data suggest that SH-J1 cells infected with Lcn2-expression adenovirus expressed Lcn2 as a 25 kDa protein, after which it was secreted into the medium where it formed homo- and heterodimeric complexes (Supporting Fig. S8). Immunohistochemical staining in our cohort of patients demonstrated that Lcn2 immunoreactivity was localized at the front of tumor tissue adjacent to the connective matrix (Fig. 1E, upper panels). Lcn2 expression increased in various tumor stages (Fig. 1E, middle panels). However, analysis of tissue microarray (TMA) data from an independent cohort of patients revealed various staining intensities according to the grade of differentiation of HCCs; very little staining was observed in dedifferentiated HCCs (GIII/IV) (Fig. 1E, lower panels). In addition, cells that had undergone EMT (SH-J1 and SCK) and EMT-relevant HCCs (GIV) appeared to express less Lcn2 than cells and tissues with an epithelial phenotype. Thus, we investigated the correlation between Lcn2 expression and EMT marker expression in HCC samples using TMA analysis (Supporting Fig. S9). The staining intensity of Lcn2 was positively correlated with the expression of epithelial markers such as E-cadherin (P < 0.001), desmoplakin I/II (P < 0.001), and CK18 (P = 0.03), and inversely correlated with the expression of fibronectin (P = 0.016) and vimentin (P = 0.002), implying that Lcn2 expression is positively correlated with epithelial marker expression and inversely related to EMT marker expression (Supporting Table S6). In our cohort of patients, statistical analysis revealed that Lcn2 immunoreactivity was positively correlated with stage, but not Edmondson differentiation grade or recurrence, in patients without a distant metastasis who underwent surgical resection (Table 1). Immunofluorescence assays revealed that GFP-tagged Lcn2 overlapped with Lcn2 immunoreactivity and that GFP-tagged Lcn2 was localized in both the nucleus and cytoplasm of Hep3B and THLE2 cells (Supporting Fig. S10), whereas endogenous Lcn2 was localized mainly in the cytoplasm.
|Edmondson-Steiner Differentiation Grade (n = 105)||TNM Stage (n = 144)||Recurrence (n = 106)|
|Score||−0.052 (0.601)a||0.194 (0.020)a||0.185 (0.058)|
Lcn2 Inhibits the Proliferation and Tumorigenicity of SH-J1 Cells With an EMT Phenotype
To determine whether human Lcn2 is involved in tumor cell proliferation or tumorigenicity in HCC cells with an EMT phenotype, we established SH-J1 cells stably expressing Lcn2. The transfectants showed a more adherent morphology than vector control cells (Fig. 2A). Furthermore, the tetrazolium salt 3-(4,5-dimethylthiazol-2-yl)−2,5-diphenyl tetrazoliumbromide (MTT) assay revealed that the proliferation rate of SH-J1 cells stably expressing Lcn2 was lower than that of vector control cells (Fig. 2B). Next, Lcn2-expressing SH-J1 cells were inoculated subcutaneously into nude mice to determine whether Lcn2 affects tumorigenicity. Tumor growth was more sluggish when stable transfectants rather than vector control cells were used for tumor induction (Fig. 2C). Vector control cells (VC1 and VC2) formed bigger tumor masses than Lcn2-expressing cells (Fig. 2D, upper panels). Lcn2 immunoreactivity was found mainly in the cytoplasm and to a lesser extent in the nuclei of tumor tissues (Fig. 2D, lower panels).
Acquisition of EMT Characteristics by HKK-2 Cells With an Epithelial Phenotype by Lcn2 Knockdown
First, using transient expression of Lcn2 by adeno-associated virus transduction, we examined whether Lcn2 influences the expression of EMT-associated markers in SH-J1 cells. Lcn2 effectively inhibited the expression of mesenchymal markers such as N-cadherin (N-cad), alpha-smooth muscle actin (α-SMA), vimentin (VIM), and fibronectin (FN), which are EMT marker genes. In contrast, Lcn2 treatment increased the expression of epithelial markers, including cytokeratin 8 (CK8), cytokeratin 18 (CK18), and desmoplakin I/II (DesI/II) (Fig. 3A). Next, we performed knockdown of Lcn2 by small hairpin RNA (shRNA) lentiviral delivery in HKK-2 cells and observed the simultaneous up-regulation of mesenchymal markers and down-regulation of epithelial markers (Fig. 3B). These results are consistent with the results we obtained by overexpressing Lcn2 in SH-J1 cells by adenovirus infection.
Growth factor signaling pathways, particularly EGF- and EGFR-driven signaling pathways, have been shown to play a crucial role in cancer progression. Overexpression of EGF and EGFR has been reported in various cancer types, including HCC.[24, 25] It has also been demonstrated that EGF treatment in vitro enhances the invasiveness and metastatic properties of several different cancer cells, including ovarian, cervical, epidermoid, and breast cancer cells. To investigate whether EGF is involved in the down-regulation of Lcn2 and the up-regulation of Twist1 in HCC and CC cells, we examined the effects of EGF on Lcn2 expression in HLK-5 and JCK cells, which strongly express Lcn2 (Fig. 3C). EGF treatment resulted in cells with a migratory and scattering phenotype and the down-regulation of Lcn2 and E-cad and up-regulation of Twist1. Furthermore, concomitant treatment of cells with the EGF receptor tyrosine kinase inhibitor, AG1478, substantially blocked these EGF-mediated changes. It has also been demonstrated that loss of E-cadherin is a causal factor that promotes tumor progression. In our study, EGF treatment remarkably reduced E-cadherin protein expression concurrent with a reduction in the protein level of Lcn2, accompanied by increased Twist1 expression. TGF-β1 treatment and EGF had similar effects on cell morphology and epithelial marker expression (Fig. 3D).
Lcn2-Induced Inhibition of Invasion and Metastasis in HCC Cells
Wound repair assays were performed using Lcn2-negative (SK-HEP1 and Huh7) or Lcn2-positive cell lines (HKK-2 and HLK-5, respectively) (Fig. 4A, left panels). The wound closure ability of Lcn2-positive cell lines was significantly lower than that of Lcn2-negative cell lines, even though the Lcn2-positive cell lines expressed much more endogenous EGF and TGF-β1 than the Lcn2-negative cell lines (Fig. 4A, middle panel). Next, we investigated whether ectopic Lcn2 expression by SH-J1 cells significantly inhibited wound closure (Fig. 4A, right panels). Cells harboring stable transfectants were significantly less likely to migrate to the wounded area compared with parent or vector control cells. In a modified Boyden chamber assay, stable transfectants penetrated the matrix and colonized the bottom surface of the Matrigel-coated membrane to a lesser extent than vector control cells; this was true for both SH-J1 (Fig. 4B, left panels) and SH-J1-luc stable transfectants (Fig. 4B, right panels). Lcn2 expression therefore appears to inhibit the migration and invasiveness of cells in vitro. We also evaluated the effects of EGF and TGF-β1 treatment on the ability of Lcn2-positive or -negative cell lines to close wounds (Supporting Fig. S11). EGF and TGF-β1 treatment significantly enhanced the wound closure ability of HCC cells endogenously expressing Lcn2, but not that of HCC cells ectopically expressing Lcn2. These results suggested that Lcn2 functions downstream of EGF and TGF-β1 signaling. Next, to determine the functional role of Lcn2 in HCC cell metastasis, we used SH-J1-luc cells expressing luciferase and established Lcn2-expressing SH-J1-luc cells (Lcn2-luc) (Fig. 4C, left panels). The metastatic phenotype of the Lcn2-luc cells was examined by injection of the cells into the tail veins (200 μL of 5 × 105 cells) of nude mice, followed by detection of multiple metastatic nodules in the lungs (Fig. 4C, right panel). Vector control cells first colonized and then continued growing in the lungs with many metastatic nodules, whereas Lcn2-expressing cells formed far fewer metastatic nodules in the lungs. Sufficient bioluminescence data were obtained 40 days postinjection (Fig. 4D). These results suggest that Lcn2 plays a critical role in inhibiting metastasis and invasion in HCC.
Lcn2 Transcriptionally Down-Regulates Twist1
Twist expression has been shown to promote migration and invasion in HCC. We found that Twist1 protein expression was significantly down-regulated in stable transfectants expressing Lcn2 (Fig. 5A) and in cells transduced with Lcn2-expressing adenovirus (Fig. 5B; Supporting Fig. S12). In contrast, Twist2, Slug, and Snail expression did not change (Supporting Fig. S13). Next, to investigate whether Twist1 down-regulation is dependent on transcriptional regulation, we examined Lcn2-mediated Twist1 mRNA expression by real-time PCR analysis in HEK293T (Supporting Fig. S14, left panel) and SH-J1 cells (Fig. 5C, left panel). We found that Twist1 down-regulation was associated with a decrease in transcript levels of this gene. Furthermore, a promoter assay revealed that Lcn2 effectively decreased Twist1 promoter activity in HEK293T (Supporting Fig. S14, right panel) and SH-J1 cells transfected with a Twist1-luc construct containing the human Twist1 promoter linked to a luciferase reporter gene (Fig. 5C, right panel). Chromatin immunoprecipitation assays were performed to evaluate in vivo Lcn2 binding to the promoter DNA of Twist1 in SH-J1 cells (Fig. 5D). Lcn2 was able to effectively bind to the promoter of Twist1 at the promoter 1 site (+15 to −179).
EMT Changes Are Dependent on the EGF (or TGF-β1)/Lcn2/Twist1 Pathway
Finally, we examined changes in EMT marker expression in the Lcn2 transfectants after treatment with TGF-β1. TGF-β1 treatment did not change EMT marker expression in the Lcn2 transfectants, whereas TGF-β1 treatment substantially increased the expression of mesenchymal markers (VIM, N-cad, and FN) and decreased that of epithelial markers (CK8 and CK18) in vector control cells (Fig. 6A). These results suggest that Lcn2 overexpression acts upstream of Twist 1 to block TGF-β1-mediated EMT changes. Next, we knocked-down Twist1 in SH-J1 cells, which have a mesenchymal phenotype and express Twist1. Twist1 knockdown resulted in decreased expression of mesenchymal markers (VIM, N-cad, and α-SMA) and increased expression of epithelial markers (CK8 and CK18) (Fig. 6B). These results suggest that Twist1 acts upstream of EMT marker expression and downstream of Lcn2. Next, we stably expressed Twist1 in HLK2 cells, which endogenously express Lcn2. Twist1 expression induced morphological changes in the cells consistent with a migratory or invasive phenotype (Fig. 6C). Furthermore, Twist1 enhanced the expression of mesenchymal markers (VIM, Nx02010;cad, FN, and α-SMA) and reduced the expression of epithelial markers (CK8 and CK18) in HLK2 cells (Fig. 6D). These results suggest that Twist1 acts downstream of Lcn2 and is the final regulator of EMT change in HCC progression (Fig. 6E).
- Top of page
- Materials and Methods
- Supporting Information
A number of recent gene expression profiling studies have identified genes specifically up-regulated or down-regulated in HCC tissues with the ultimate goal of developing novel diagnostic markers for early detection, prognostic markers for prediction of clinical outcome, and therapeutic targets for tumor progression intervention. Previous studies identified Lcn2 as a gene that is highly up-regulated in HCC tissues.[12-14] Although the mechanism regulating the expression of Lcn2 in HCC cells in vivo is not fully understood, inflammatory cytokine interleukin (IL)−1β induces Lcn2 expression in Huh7 cells, in primary rat hepatocytes, and in human epithelial cells. In addition, chronic liver inflammation and hepatic regeneration, induced in part by infection with hepatitis B or hepatitis C virus, and the consequent cellular immune responses, may increase the risk of HCC development by favoring the accumulation of genetic alterations in hepatocytes that might trigger specific oncogenic pathways. We demonstrated that Lcn2 mRNA and protein levels were higher in HCC tissues than in corresponding nontumor tissues. Further cluster analysis of subgroups revealed that this differential expression pattern resulted from differences in expression between LC and GI/II HCC samples. Interestingly, our data showed that SH-J1 and SCK cells with an EMT phenotype barely expressed Lcn2. Some epithelial HCC cells (HLK-2, HKK-2, and HLK-5) and epithelial cholangiocarcinoma cells (Choi-CK, Cho-CK, and JCK) show moderate to strong expression of Lcn2. Lcn2 was also reported to be a biomarker in cholangiocarcinoma; these findings suggest that the role of Lcn2 is dependent on the liver tumor with an epithelial phenotype. Our results are in good agreement with previous observations in cancer of the ovary, where Lcn2 expression is almost completely absent in tissue samples from normal ovaries, but strongly expressed in both borderline and grade 1 tumors, and weakly to moderately expressed in grade 2 and 3 tumors. Furthermore, moderate to strong expression of Lcn2 was observed in the epithelial ovarian cancer cell lines SKOV3 and OVCA433, while no expression was detected in mesenchymal-like OVHS1, PEO36, or HEY cell lines. Taken together, Lcn2 expression appears to be linked to the epithelial phenotype of tumors and is lost as the tumor progresses and becomes undifferentiated.
Lcn2 has been implicated in the induction of cellular proliferation because its expression is associated with a variety of proliferative cells.[36, 37] Lcn2 expression is required for Bcr-Abl-induced tumorigenesis in leukemia cells. Lcn2 expression promotes breast tumor growth and progression,[39-41] and also increases colon cancer migration and invasion. Furthermore, down-regulation of Lcn2 by antisense RNA suppresses human esophageal carcinoma SHEEC cell invasion in vivo by reducing matrix metalloproteinase (MMP)−9 activity. In the present study, Lcn2 suppressed the proliferation, migration, and invasion of EMT phenotypic HCC cells in vitro and tumor growth and metastasis in vivo. This is consistent with results from a previous study that reported that Lcn2 suppressed Ras-transformed 4T1 mouse mammary tumor cell invasiveness in vitro and tumor growth and lung metastases in vivo. Furthermore, Lcn2 blocked human colon cancer KM12SM cell invasion and liver metastasis. A recent study also proposed that Lcn2 may act as a suppressor of invasion and angiogenesis in advanced pancreatic cancer cells. These apparently conflicting observations could be due to distinct functions of Lcn2 in different cell types. Our focus in this study was to determine the mechanisms by which Lcn2 inhibits growth factor-mediated EMT in association with invasion and metastasis in HCC. Loss of E-cadherin expression has been associated with activation of the EGF/EGFR cascade in several cancer types, including pancreatic cancer and cervical cancer.[25, 27] EGF-induced EMT phenotypes were inhibited in the presence of AG1478, an inhibitor of EGF receptor tyrosine kinase activity. Lim et al. reported that EGF down-regulated both E-cadherin and Lcn2 expression in ovarian cancer cells. Although the detailed mechanism of Lcn2-mediated E-cadherin regulation is not clear, these data indicate that Lcn2 may be a good marker to monitor the transition from a benign to a premalignant or malignant tumor and that Lcn2 may be involved in the progression of epithelial malignancies. Similarly, we found that Lcn2 expression was positively correlated with a worse HCC differentiation grade before dedifferentiation. Thus, Lcn2 may also be a molecular marker for the progression of HCC before tumor metastasis or EMT.
In HCC, the up-regulation and nuclear relocation of the EMT regulator Twist1 have been implicated in tumor invasion and metastasis.[31, 44] As a major regulator of EMT-mediated invasion and metastasis, Twist1 plays an important role through its regulation of E-cadherin expression, which is believed to be critical for tumor invasion. It is widely accepted that loss of E-cadherin plays a critical role in the EMT, an early event in cancer cell invasion and metastasis. In our study, the effects of Lcn2 on cell invasion and metastasis were mediated through suppression of the transcription factor Twist1 and subsequent up-regulation of E-cadherin. We found that Lcn2 can effectively translocate to the nucleus from the cytoplasm and bind to the promoter region of Twist1, which could result in the transcriptional down-regulation of Twist1. We also found that in HCC cells, EGF- or TGF-β1-mediated EMT resulted from the down-regulation of Lcn2 and subsequent up-regulation of Twist1.
In conclusion, Lcn2 inhibits proliferation, invasion, and metastasis in vitro and in vivo through transcriptional suppression of Twist1 in HCC cells. Thus, Lcn2 may be a candidate metastasis suppressor due to its ability to reverse the EMT (MET) in HCC. Lcn2 is therefore a potential therapeutic target in HCC.
- Top of page
- Materials and Methods
- Supporting Information
We thank Jack B. Cowland for providing the reporter plasmids.
- Top of page
- Materials and Methods
- Supporting Information
- 4Neutrophil gelatinase-associated lipocalin (NGAL) an early-screening biomarker for ovarian cancer: NGAL is associated with epidermal growth factor-induced epithelio-mesenchymal transition. Int J Cancer 2007;120:2426-2434., , , , , , et al.
- 35Proteomic studies of cholangiocarcinoma and hepatocellular carcinoma cell secretomes. J Biomed Biotechnol 2010;2010:437143., , , , , , et al.
- Top of page
- Materials and Methods
- Supporting Information
Additional Supporting Information may be found in the online version of this article.
|hep26467-sup-0002-suppfig1.eps||912K||Supplementary Figure 1.|
|hep26467-sup-0003-suppfig2.eps||713K||Supplementary Figure 2.|
|hep26467-sup-0004-suppfig3.eps||2488K||Supplementary Figure 3.|
|hep26467-sup-0005-suppfig4.eps||902K||Supplementary Figure 4.|
|hep26467-sup-0006-suppfig5.eps||647K||Supplementary Figure 5.|
|hep26467-sup-0007-suppfig6.eps||401K||Supplementary Figure 6.|
|hep26467-sup-0008-suppfig7.eps||544K||Supplementary Figure 7.|
|hep26467-sup-0009-suppfig8.eps||827K||Supplementary Figure 8.|
|hep26467-sup-0010-suppfig9.eps||3112K||Supplementary Figure 9.|
|hep26467-sup-0011-suppfig10.eps||821K||Supplementary Figure 10.|
|hep26467-sup-0012-suppfig11.eps||1388K||Supplementary Figure 11.|
|hep26467-sup-0013-suppfig12.eps||778K||Supplementary Figure 12.|
|hep26467-sup-0014-suppfig13.eps||525K||Supplementary Figure 13.|
|hep26467-sup-0015-suppfig14.eps||653K||Supplementary Figure 14.|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.