MicroRNA-140-5p suppresses tumor growth and metastasis by targeting transforming growth factor β receptor 1 and fibroblast growth factor 9 in hepatocellular carcinoma

Authors


  • Potential conflict of interest: Nothing to report.

  • Supported by the National Keystone Basic Research Program of China (No. 2009CB521801), National High Technology Research and Development Program of China (No. 2006AA02Z4B2), Clinical Subjects' Key Project of Ministry of Health (2007-2009, 2010-2012), and National Science & Technology Major Projects (2009ZX09103-681, 2012ZX10002012-011), and National Nature Science Foundation of China (No. 81172018).

Abstract

By comparing the expression profiles of microRNAs (miRNAs) in different hepatocellular carcinoma (HCC) subtypes, we identified miR-140-5p as an HCC-related miRNA. We found that miR-140-5p was significantly decreased in HCC tissues and all of six liver cancer cell lines examined and its expression levels were correlated with multiple nodules, vein invasion, capsular formation, and differentiation, as well as overall and disease-free survival of HCC. We also found that miR-140-5p suppressed HCC cell proliferation and HCC metastasis. Multipathway reporter arrays suggested that miR-140-5p inhibited transforming growth factor β (TGF-β) and mitogen-activated protein kinase / extracellular signal-regulated kinase (MAPK/ERK) signaling. TGFB receptor 1 (TGFBR1) and fibroblast growth factor 9 (FGF9) were then characterized as the direct targets for miR-140-5p after it was found that ectopic miR-140-5p expression suppressed TGFBR1 and FGF9 expression. Silencing TGFBR1 and FGF9 by small interfering RNA (siRNA) resembled the phenotype resulting from ectopic miR-140-5p expression, while overexpression of TGFBR1 and FGF9 attenuated the effect of miR-140-5p on HCC growth and metastasis. Conclusion: These data elucidated a tumor suppressor role for miR-140-5p in HCC development and progression with therapeutic potential. Our correlation studies in clinical HCC samples further suggest that miR-140-5p could be a valuable biomarker for HCC prognosis. (Hepatology 2013 )

Hepatocellular carcinoma (HCC) is one of the most common human cancers in the world, particularly in China.1 It ranks as the fifth most common malignancy and the second leading cause of cancer death worldwide, resulting in more than 695,900 deaths each year.2, 3 Although its mortality decreased along with advances in surgical resection, the long-term prognosis remains unsatisfactory. For example, the 5-year survival rate is only 20% to 30% in HCC patients after surgical resection, mainly due to the high recurrence and metastasis rate.4, 5 It has been generally accepted that the invasive and metastatic potentials of HCC are mostly attributed to the differences of pathological and molecular characteristics.6 Previously, we found a specific subtype of HCC in which the tumor was only around 5 cm in diameter with a single lesion, but the tumor grew expansively within an intact capsule or pseudocapsule. More important, the tumor possessed unique clinical and pathological characteristics. Therefore, we categorized it as solitary large hepatocellular carcinoma (SLHCC) and divided HCC into three different subtypes: SLHCC, nodular HCC (NHCC, node number ≥2), and small HCC (SHCC, tumor ≤5 cm). Further study confirmed that the metastatic potential of SLHCC was comparable to SHCC, but significantly less than NHCC.5 Additionally, SLHCC exhibited a similar long-term overall and disease-free survival to SHCC after hepatic resection, but much better than NHCC.5 Although much work had been done,7-11 the exact mechanisms that determine the differences for molecular characteristics and metastatic potential among these three subtypes of HCC are still elusive.

MicroRNAs (miRNAs) are a class of small, endogenously expressed, well-conserved noncoding RNA molecules with 18-25 nucleotides (nt). They play important regulatory roles by targeting messenger RNAs (mRNAs) for cleavage or translational repression.12 Given that more than 50% of miRNAs are located in cancer-associated genomic regions or in fragile sites, miRNAs may play an important role in cancer pathogenesis.13 Indeed, aberrant miRNA expression has been demonstrated in HCC, which contributes to carcinogenesis and cancer development by promoting oncogene expression or by inhibiting tumor suppressor genes.14-17

To further demonstrate whether miRNAs can serve as promising prognostic markers for HCC and to discover their implication in HCC invasion and metastasis, we profiled miRNA expression by analysis of 840 mammalian miRNAs in HCC from 30 HCC samples. miR-140-5p was identified to be significantly down-regulated in HCC tissues as compared with that of adjacent nontumorous liver tissues (ANLTs). However, miR-140-5p displayed similar expression levels between SLHCC and SHCC, but much lower levels of miR-140-5p was noted in NHCC. These results were further confirmed by quantitative real-time polymerase chain reaction (PCR) in 120 paired cases of human HCC tissues and cell lines. Functional studies revealed that miR-140-5p targets transforming growth factor β receptor 1 (TGFBR1) and fibroblast growth factor 9 (FGF9), by which it suppresses HCC cell proliferation and metastasis.

Abbreviations

FGF9, fibroblast growth factor 9; HCC, hepatocellular carcinoma; NHCC, nodular hepatocellular carcinoma; qRT-PCR, real-time quantitative PCR; SHCC, small hepatocellular carcinoma; SLHCC, solitary large hepatocellular carcinoma; TGFR1, transforming growth factor β receptor 1.

Materials and Methods

Patients and Tissue Specimens.

Matched fresh HCC specimens and ANLTs were obtained from 120 patients during hepatic resection at the Department of Surgery, Xiangya Hospital of Central South University (CSU) from January 2004 to October 2007. Details are described in the Supporting Materials. Prior informed consent was obtained and the study protocol was approved by the Ethics Committee of Xiangya Hospital of CSU.

miRNA Array.

miRCURY LNA microRNA chips (v. 8.0, Exiqon, Vedbaek, Denmark) were used to profile the differences for miRNA expression among SLHCC, SHCC, and NHCC. The array contained a total of 840 specific probes in triplicate. It was performed according to the protocol of miRCURY LNA microRNA Array Power Labeling kit (Exiqon).18 The image analysis was conducted in Genepix Pro 6.0 (Axon Instruments) as described19 (details in the Supporting Materials and Methods).

Cell Lines and Cell Culture.

HCCLM3 and MHCC97-L cell lines were used for this study. Details are described in the Supporting Materials and Methods.

Quantitative Real-Time PCR (qRT-PCR).

qRT-PCR was performed using the TaqMan MicroRNA reverse transcription kit and TaqMan Universal PCR Master Mix (Ambion, Austin, TX). Details are described in the Supporting Materials and Methods.

Western Blot Analysis.

Details are described in the Supporting Materials and Methods.

Follow-up and Prognostic Study.

Follow-up data were obtained after hepatic resection for all 120 patients. Details are described in the Supporting Materials.

Vector Construction and Transfection.

The DNA fragment for miR-140-5p was amplified from genomic DNA and inserted into Age I / EcoR I site of a lentiviral expression vector pGCSIL-GFP (GeneChem, Shanghai, China). The TGFBR1 and FGF9 expression vectors were constructed by inserting their ORF sequence into the pGCL vector (GeneChem, Shanghai, China). siRNAs were purchased from GenePharma (Shanghai, China) and the sequences of these siRNAs are provided in Supporting Table 1. Transfection was performed according to the manufacturer's protocol. Viruses were harvested 72 hours after transfection and viral titers were 1 × 109 TU/mL; 1 × 105 cells were infected with 2 × 106 lentivirus in the presence of 6 μg/mL polybrene (Sigma, St. Louis, MO). In the present study, the infection efficiency of lentivirus was over 90% (Supporting Fig 1). No significant cell death was observed after virus infection. Bulk transfectants were used for subsequent assays. For luciferase analysis, the 3′-untranslated region (UTR) sequence of TGFBR1 and FGF9 were amplified from human liver genomic DNA and then cloned into the downstream region of a firefly luciferase cassette in the pGL3-Promoter vector (Promega, Madison, WI) as instructed.

Cell Proliferation, Cell Cycle Analysis, and Colony Formation Assays.

Details are described in the Supporting Materials and Methods.

In Vitro Wound Healing (Migration) and Invasion Assays.

Details are described in the Supporting Materials and Methods.

Multipathway Reporter Array.

ACignal Finder 10-Pathway Reporter Array (SABiosciences) was employed for the study. A reverse transfection technique was implemented. Cells were treated with overexpression miR-140-5p or negative control. Relative firefly luciferase activity was calculated and normalized to the constitutively expressed Renilla luciferase.

Luciferase Reporter Assay.

Luciferase activity was assessed according to the Dual-Luciferase Reporter Assay protocol (Promega, Madison, WI) using a Veritas 96-well Microplate Luminometer (Promega) with substrate dispenser (Promega). HEK293T cells transduced with leti-miR-140-5p or control virus were seeded in 96-well plates with 70% confluence. Twelve hours later, the cells were cotransfected with 50 ng pGL3-Promoter -UTR and 10 ng pRLTK using Lipofectamine LTX. After 24 hours of transfection, the cells were harvested for firefly and Renilla luciferase activity assay. The Renilla luciferase activities were used to normalize the transfection efficiency.

HCC Mouse Model.

The HCC model in nude mice was constructed as described.26 Details are described in the Supporting Materials and Methods.

Immunohistochemistry.

The expression levels for TGFBR1 and FGF9 in the local tumor tissues were determined by immunostaining with antibodies against TGFBR1 and FGF9 (Santa Cruz Biotechnology, Santa Cruz, CA). All animal studies were conducted at the Animal Institute of CSU according to the protocols approved by the Medical Experimental Animal Care Commission of CSU.

Statistical Analysis.

Statistical analysis was performed using SPSS (v. 13.0, Chicago, IL). Data for miR-140-5p expression in fresh specimens were analyzed using the Mann-Whitney U test. The Fisher's exact test was used for statistical analysis of categorical data. A Spearman correlation test was used for analyzing the correlations between miR-140-5p expression level and the clinical and pathological variables. Survival curves were constructed using the Kaplan-Meier method and evaluated using the log-rank test. The Cox proportional hazard regression model was used to identify factors that were independently associated with overall survival and disease-free survival. P < 0.05 was considered statistically significant.

Results

Down-Regulation of miR-140-5p Is Associated with HCC Poor Prognosis.

Our miRNA microarray analysis revealed that miR-140-5p was significantly down-regulated in HCC tissues (Fig. 1A). To confirm this result, we performed qRT-PCR in 120 cases of HCC tissues and ANLTs. In general, a 3.4-fold decrease for miR-140-5p expression was noted in HCC tissues as compared with that of ANLTs (Fig. 1B). Comparative analysis of paired HCCs with ANLTs further revealed that reduced miR-140-5p expression (more than 2-fold [i.e., log2 (fold change) > 1]) was observed in 89 (74.2%) cases, suggesting that reduction of miR-140-5p was a frequent event in human HCC (Fig. 1B). We also measured miR-140-5p expression in six liver cancer cells (SMMC7721, HepG2, MHCC97-L, MHCC97-H, HCCLM3, and Huh7), and normal L02 liver cells were used as a control. Similarly, the relative expression levels for miR-140-5p in these six HCC cells were 0.41, 0.48, 0.45, 0.40, 0.31, and 0.36 as compared with that of L02 cells, respectively (Fig. 1E). Next, analysis was conducted between two groups (60 cases for each group), one with high miR-140-5p expression, while the other with low miR-140-5p expression (cutoff value of median). It was found that miR-140-5p expression was significantly associated with tumor nodule number (P = 0.036), capsular formation (P = 0.002) and TNM stage (P = 0.001) (Table 1).

Figure 1.

miR-140-5p is frequently down-regulated in HCC and is a promising prognostic marker for HCC. (A) Cluster analysis of miRNAs expression profiles of SHCC, SLHCC, and NHCC versus ANLTs. Overexpression is indicated in red, whereas underexpression is coded in green. (B) Expression of miR-140-5p in 120 pairs of HCC tissues and the corresponding ANLTs. Expression levels of miR-140-5p were normalized to the corresponding levels of U6 snRNA. Data were analyzed using a ΔΔCt approach and expressed as log2 fold change (ΔΔCt [HCC/ANLT]). (C,D) Survival relevance analysis of miR-140-5p expression in HCCs. According to the qRT-PCR data, the expression of miR-140-5p was classified into low expression (n = 60) and high expression (n = 60). (E) miR-140-5p expression in six liver cancer cell lines and normal liver cell line (L02). Expression levels of miR-140-5p were normalized to the corresponding levels of U6 snRNA. Each sample was analyzed in triplicate and values are expressed as levels (mean ± standard deviation) relative to those in L02 cells. (F) miR-140-5p expression in SHCC, SLHCC, and NHCC was calculated by a ΔΔCt approach and expressed as log2 fold change (ΔΔCt [HCC/ANLT]). The expression level was compared with an independent t test. (G,H) Survival analysis of three subtypes of HCC. According to the clinical data, 120 HCC patients were divided into three subtypes (SHCC, SLHCC, and NHCC). Survival curves were constructed using the Kaplan-Meier method and evaluated using the log-rank test.

Table 1. Correlations Between mir-140-5p Expression Level and Clinicopathological Variables of 120 Cases of HCC
  miR-140-5p Expression
Clinicopathologic VariablesnLowHighP
Gender    
  Male1045054 
  Female16970.547
Age (years)    
  ≤60974552 
  >60231490.215
Cirrhosis    
  Absence451926 
  Presence7540350.242
Tumor size (cm)    
  ≤5421725 
  >57843350.128
Tumor nodule number    
  Solitary773344 
  Multiple (≥2)4327160.036
Capsular formation    
  Presence531835 
  Absence6943260.002
Edmondson-Steiner grade    
  I-II612426 
  III-IV5921190.676
Vein invasion    
  Presence642935 
  Absence5630260.371
 TNM stage    
  Stage I21417 
  Stage II512328 
  Stage III4830180.001
Liver function    
  Child-Pugh A1085355 
  Child-Pugh B12750.547

To assess the feasibility of miR-140-5p expression in HCC prognosis, the Cox proportional hazards regression model was introduced. On univariate survival analysis, cirrhosis (P = 0.001), tumor nodule number (P = 0.001), vein invasion (P < 0.001), TNM stage (P < 0.001), liver function (P = 0.046), and miR-140-5p expression (P = 0.013) reached significance for overall survival. Next, multivariate survival analysis on all parameters was performed. We found that overall survival time was significantly dependent on cirrhosis (P = 0.001), vein invasion (P = 0.004), and miR-140-5p expression levels (P = 0.022; Table 2). Furthermore, HCC patients with high miR-140-5p expression had much longer overall survival time (median survival time, 36.7 versus 14.3 months, P = 0.011) than those with low miR-140-5p expression (Fig. 1C). For analysis of disease-free survival time, the tumor nodule number (P = 0.030), vein invasion (P = 0.012), TNM stage (P = 0.013) and miR-140-5p expression (P = 0.002) reached significance in the univariate Cox proportional hazards regression model. In contrast, only the tumor nodule number (P = 0.032), vein invasion (P = 0.014), and miR-140-5p expression (P = 0.013; Table 3) reached significance for disease-free survival time on multivariate survival analysis. Similarly, HCC patients with high miR-140-5p expression had longer disease-free survival (median survival time, 34.0 versus 14.0 months, P = 0.006; Fig. 1D) than those with low miR-140-5p expression. To determine whether a lower level of miR-140-5p is associated with higher metastasis rates, we examined the miR-140-5p in 40 patients with intrahepatic metastasis or lung metastasis. Intrahepatic metastasis and lung metastasis were found in 25 and 2 patients with lower levels of miR-140-5p, respectively. In contrast, intrahepatic metastasis and lung metastasis were found in 13 and 0 patients with higher levels of miR-140-5p, respectively.

Table 2. Cox Regression Analyses of Overall Survival (OS) and mir-140-5p Expression Level as Well as Clinicopathological Parameters
  Univariate AnalysisMultivariate Analysis
VariablesnHR (95% CI)PHR (95% CI)P
Gender     
  Male1041   
  Female161.12 (0.48-2.63)0.797NANA
Age (years)     
  ≤60971   
  >60231.74(0.88-3.44)0.110NANA
Cirrhosis     
  Absence451 1 
  Presence755.97(2.14-16.66)0.0015.51 (1.93-15.70)0.001
Tumor size (cm)     
  ≤5421   
  >5780.83 (0.47-1.46)0.560NANA
Tumor nodule number     
  Solitary771 1 
  Multiple (≥2)432.54 (1.45-4.45)0.0011.94 (0.92 -4.07)0.081
Capsular formation     
  Presence531   
  Absence691.16 (0.66-2.04)0.264NANA
Edmondson-Steiner grade     
  I-II611   
  III-IV591.59 (0.91-2.79)0.103NANA
Vein invasion     
  Presence641 1 
  Absence563.44 (1.89-6.25)<0.0013.52 (1.49-8.32)0.004
TNM stage     
  Stage I211 1 
  Stage II51    
  Stage III482.31 (1.45-3.68)<0.0010.87 (0.42-1.79)0.698
Liver function     
  Child-Pugh A1081 1 
  Child-Pugh B122.41 (1.02-5.73)0.0460.95 (0.36-2.48)0.916
MiR-140-5p expression     
  Low601 1 
  High602.07 (1.16-3.68)0.0132.13 (1.11-4.07)0.022
Table 3. Cox Regression Analyses of Disease-Free Survival (DFS) and mir-140-5p Expression Level as Well as Clinicopathological Parameters
  Univariate AnalysisMultivariate Analysis
VariablesnHR (95% CI)PHR (95% CI)P
Gender     
  Male1041 1 
 Female161.11 (0.50-2.49)0.7931.11 (0.43-2.88)0.835
Age (years)     
  ≤60971 1 
  >60231.86 (0.94-3.69)0.0751.68(0.83-3.38)0.150
Cirrhosis     
  Absence451 1 
  Presence751.70 (0.90-3.23)0.1031.78 (0.90-3.60)0.098
Tumor size (cm)     
  ≤5421 1 
  >5781.41 (0.77-2.58)0.2640.83 (0.23-3.04)0.834
Tumor nodule number     
  Solitary771 1 
  Multiple (≥2)431.86 (1.06-3.25)0.0301.93 (1.06-3.51)0.032
Capsular formation     
  Presence531 1 
  Absence691.74 (0.98-3.09)0.0591.14 (0.56-2.36)0.715
Edmondson-Steiner grade     
  I-II611 1 
  III-IV591.36 (0.78-2.37)0.2831.69 (0.90-3.16)0.104
Vein invasion     
  Presence641 1 
  Absence561.97 (1.31-3.10)0.0121.99 (1.15-3.44)0.014
TNM stage     
  Stage I211 1 
  Stage II51    
  Stage III481.69 (1.12-2.56)0.0130.91 (0.46-1.79)0.778
Liver function     
  Child-Pugh A1081 1 
  Child-Pugh B122.26 (0.89-5.77)0.0880.96 (0.26-3.53)0.949
MiR-140-5p expression     
  Low601 1 
  High602.43 (1.37-4.31)0.0022.42 (1.21-4.86)0.013

Higher Levels of miR-140-5p Expression Are Associated With Longer HCC Survival and Reduced Tumor Recurrence.

We further analyzed the expression differences of miR-140-5p in three subtypes of HCC (SHCC, SLHCC, and NHCC). Consistent with the miRNA array data, similar median miR-140-5p expression levels were noticed between SLHCC and SHCC (0.35 versus 0.375, P = 0.935; Fig. 1F), but significantly lower levels of miR-140-5p were noted in NHCC (0.35 versus 0.2, P < 0.001; Fig. 1F). Remarkably, higher levels of miR-140-5p expression were associated with longer survival time. For example, SHCC and SLHCC had longer survival time than NHCC and the medium overall cumulative survival of SHCC, SLHCC, and NHCC were 56.7, 33.3, and 15.3 months, respectively (P = 0.004; Fig. 1G). In line with this result, the median disease-free survival of SLHCC was 28.0 months, which was significantly better than that of NHCC (14.0 months), but similar to that of SHCC (38.0 months, P = 0.003; Fig. 1H). These results indicated that miR-140-5p might play a critical role in HCC metastasis and progression.

miR-140-5p Inhibits HCC Growth and Metastasis.

To determine the biological significance of miR-140-5p in HCC metastasis, we performed a wound-healing assay and transwell assay using HCCLM3 and MHCC97-H cells. It was noted that ectopic miR-140-5p expression significantly suppressed wound healing in the studies with both HCCLM3 and MHCC97-H cells (P < 0.01; Fig. 2A). Transwell assays with Matrigel further demonstrated that miR-140-5p markedly inhibited the invasive capacity of HCCLM3 and MHCC97-H cells as compared with that of vector-transduced control cells (P < 0.001; Fig. 2B). To further confirm these results, we next examined miR-140-5p-induced cytoskeletal and morphologic changes in HCCLM3 cells. As shown in Fig. 2C, miR-140-5p stimulated the reorganization of F-actin leading to the formation of stress fiber-like structures, while these structures were absent in vector-transduced cells.

Figure 2.

miR-140-5p inhibits HCC cell migration, invasion, growth, and colony formation in vitro. (A,B) The wound-healing assay (A) and invasion assay (B) of HCCLM3 and MHCC97-H cells infected with miR-140-5p lentivirus or control vector. The invasion assay was measured by way of transwell assays with Matrigel. (C) Immunofluorescence assay of HCCLM3 cells infected with miR-140-5p lentivirus or control vector. The F-actin filaments were visualized in cells using actin tracker green. (D) The growth of HCCLM3 and MHCC97-H cells with miR-140-5p overexpression or vector control was determined as described in Materials and Methods. *P < 0.01. (E) The cell cycle distribution of HCCLM3 cells infected with miR-140-5p lentivirus or control vector were analyzed as described in Materials and Methods. (F) The colony formation assay was performed as described in Materials and Methods. The number of colonies was counted and compared.

To demonstrate the effect of miR-140-5p on HCC growth, we performed an HCC cell proliferation assay. As shown in Fig. 2D, lentiviral-induced ectopic miR-140-5p resulted in a significant decrease in cell proliferation in both HCCLM3 and MHCC97-H cells (Fig. 2D). We then performed cell cycle analysis and revealed that miR-140-5p arrested the cells at S phase. Ectopic miR-140-5p expression decreased the percentage of cells in G1 phase from 60.92% to 38.64% (P < 0.001), but increased the percentage of cells in S phase and G2/M phase from 34.82% to 53.43% (P < 0.001) and 4.26% to 7.93% (P > 0.05), respectively (Fig. 2E). Consistent with these results, ectopic miR-140-5p expression suppressed colony formation (Fig. 2F).

To confirm the above data in vivo, we did xenotransplantation of tumor grafts. Consistently, miR-140-5p significantly inhibited tumor growth in vivo. The size of subcutaneous tumors and local liver tumors originated from miR-140-5p-transduced HCCLM3 cells were dramatically smaller than that of vector-transduced cells (P = 0.022, P = 0.013, respectively; Fig. 3A). We further examined the mice for liver and lung metastasis of the carcinoma cells. As shown in Fig. 3B, the intrahepatic metastasis rates in mice with miR-140-5p-transduced grafts was only 20%, while metastasis was completely absent in the lung. In contrast, mice engrafted with vector-transduced tumors showed 80% and 60% rates for intrahepatic and lung metastasis, respectively. Taken together, our data support the conclusion that miR-140-5p suppresses HCC growth and metastasis.

Figure 3.

miR-140-5p inhibits HCC cell growth and metastasis in vivo. (A) The HCC mouse model was constructed using HCCLM3 cells infected with control vector or miR-140-5p lentivirus. The size of liver tumors in these two groups was calculated and compared. (B) Intrahepatic and lung metastasis of HCCLM3 cells. Hematoxylin and eosin (H&E) stain, original magnification: ×100, ×400. (C) The percentage of mice with or without metastatic nodules in the livers or in the lungs was calculated and compared.

TGFBR1 and FGF9 Are Candidate Targets for miR-140-5p.

To dissect the mechanisms by which miR-140-5p suppresses HCC growth and metastasis, we next searched candidate targets for miR-140-5p in genes involved in HCC pathogenesis. We first used a multipathway reporter array to explore the potential signaling pathway of miR-140-5p regulated. As shown in Fig. 4A, miR-140-5p expression attenuated the activity of TGF-β and mitogen-activated protein kinase / extracellular signal-regulated kinase (MAPK/ERK) signaling, both of which are crucial for the regulation of cell migration.20-22 We therefore focused on these two pathways to search for potential targets based on those genes with oncogenic properties using the miRanda, TargetScan, and PicTar algorithms, and only those targets detected by all programs were considered. Interestingly, TGFBR1 and FGF9 were found to be the direct downstream targets, and they are implicated in TGF-β and MAPK/ERK signaling, respectively. To demonstrate that miR-140-5p binds to the 3′-UTR of TGFBR1 and FGF9, we performed miR-140-5p-based luciferase assay using the constructs described in Fig. 4B. As expected, miR-140-5p directly bound to TGFBR1 and FGF9 3′-UTR, and by which it remarkably reduced luciferase activities, whereas cells with mutant TGFBR1 and FGF9 3′-UTR displayed much higher luciferase activities (Fig. 4C). Moreover, western blot analysis and immunostaining further demonstrated that ectopic miR-140-5p dramatically suppressed the endogenous protein levels for TGFBR1 and FGF9 in HCCLM3 and MHCC97-H cells (Fig. 4D,E). Consistent with these results, attenuated expression for Smad3, p-ERK, and H-Ras were noted in miR-140-5p-transduced cells (Fig. 4D,E). Taken together, these results indicated that TGFBR1 and FGF9 were direct downstream targets for miR-140-5p in HCC cells.

Figure 4.

TGF-β R1 and FGF9 are direct downstream targets for miR-140-5p. (A) Multipathway reporter arrays were used as described in Materials and Methods for searching the possible signaling pathway of miR-140-5p. (B) miR-140-5p and its putative binding sequence in the 3′-UTR of TGF-β R1 and FGF9. The mutant miR-140-5p-binding site was generated in the complementary site for the seed region of miR-140-5p. (C) Relative luciferase activity was analyzed. HCCLM3 cells were cotransfected with pGL3-Promoter (vector) or pGL3-Promoter -miR-140-5p, firefly luciferase reporter containing either a wildtype or a mutant 3′-UTR (indicated as WT or Mut on the X axis), and a Renilla luciferase expressing construct (as internal control to calibrate the differences in both transfection and harvest efficiencies). The firefly luciferase activity of each sample was normalized to the Renilla luciferase activity. The normalized luciferase activity of wildtype pGL3-transfectants in each experiment was set as relative luciferase activity. (D) Western blot results of endogenous TGF-β R1 and FGF9 proteins in HCCLM3 cells infected with miR-140-5p lentivirus or vector control. (E) Analysis of TGF-β R1 and FGF9 expression in orthotopic implant primary tumors by immunohistochemistry.

miR-140-5p Represses TGFBR1 and FGF9 Signaling to Suppress HCC Growth and Metastasis.

The above results prompted us to examine whether miR-140-5p suppresses HCC growth and metastasis through repression of TGFBR1 and FGF9 signaling. For this purpose, we first examined whether blockage of TGFBR1 and FGF9 would mimic the effect of miR-140-5p expression. We introduced siRNA for TGFBR1, FGF9, and both TGFBR1 and FGF9 into HCCLM3 cells. Western blot analysis confirmed that the expression of TGFBR1 and FGF9 was inhibited (Supporting Fig. 2). As expected, compared to the control group, HCCLM3 cells transfected with TGFBR1 and FGF9 siRNA displayed poor wound healing (Fig. 5B) and suppressed invasive activity (Fig. 5C). Interestingly, cell proliferation assay (Fig. 5D), cell cycle analysis (Fig. 5E), and colony formation assay (Fig. 5F) confirmed that HCCLM3 cells treated with FGF9 siRNA resembled the effect of ectopic miR-140-5p expression on HCCLM3 cells, and importantly, this phenotype was not produced in cells transfected with TGFBR1 siRNA alone. Nevertheless, ectopic TGFBR1 and FGF9 expression in miR-140-5p-transduced cells attenuated the inhibitory effect of miR-140-5p on HCC growth and metastasis (Fig. 5B,C), while the inhibitory effect of miR-140-5p on HCC cell proliferation was abrogated only by overexpression of FGF9 (Fig. 5D-F). In addition, the effect of recombinant FGF9 protein is also checked. Our results showed that 1 ng/mL recombinant FGF9 protein recovered the inhibition of wound healing, invasion, and proliferation of HCC cells by miR140-5p. (Supporting Fig 3). We also examined the protein levels of TGFb1 and FGF9 receptors (FGFR2 and FGFR3). The results showed that the levels of these proteins in cells transfected with the miR-140-5p construct are the same as those in cells transfected with the control plasmid (Supporting Fig 4). In addition, knockdown of FGFR2, FGFR3, and TGFb1 were also tested. Our data show that knockdown of FGF9 receptors inhibited the invasion and proliferation of HCCLM3 cells, while knockdown of TGFb1 just inhibited the invasion of HCCLM3 cells (Supporting Figs. 5, 6).

Figure 5.

Both gain- and loss-of-function studies showed that both TGF-β R1 and FGF9 abrogate the suppressive roles of miR-140-5p in HCC cell migration and invasion and only FGF9 partly recovered the activity of proliferation of miR140-5p-transfected cells. HCCLM3 cells stably expressing miR-140-5p, or vector, were infected with or without TGF-β R1 and FGF9 lentivirus and siRNA plasmids. (A) Western blot was performed to assess the overexpressed or inhibit efficiency. Wound-healing assay (B), invasion assays (C), cell proliferation analysis (D), cell cycle distribution assay (E), and colony formation assay (F) were performed with the above cells as described in Materials and Methods to examine the effect on the biological characteristics of HCC cells. (G) Schematic representation of the major molecular mechanism of miR-140-5p suppresses tumor proliferation and metastasis.

To determine whether TGFBR1 and FGF9 regulate each other, we overexpressed TGFBR1 or FGF9 in HCCLM3 cells expressing miR-140-5p. Western blot analysis showed that the expression of the endogenous FGF9 were up-regulated by overexpression of TGFBR1 in HCC cells expressing miR-140-5p (Supporting Fig. 7A). In contrast, the expression of endogenous TGFBR1 was not affected by overexpression of FGF9 (Supporting Fig. 7B). Moreover, TGFBR1-induced invasion of HCCLM3 cells was blocked by the FGF9 siRNA (Supporting Fig. 7C,D). Our data indicate that TGFBR1 is upstream of FGF9. Taken together, our data suggest that miR-140-5p suppresses tumor invasion and metastasis by targeting TGFBR1 and FGF9, and suppresses tumor proliferation by repressing FGF9 expression.

Discussion

It is well known that each subtype of HCC exhibits distinct clinicopathological and molecular characteristics.6 Previously, we defined a specific subtype of HCC termed SLHCC.5, 10 Interestingly, although SLHCC is larger in size, it showed similar outcomes as SHCC. Both of them are better than NHCC in terms of outcomes. Our findings do not support the concept that large HCCs cannot be resected. According to this finding, many patients with SLHCC have been cured.5 Therefore, clarification of the molecular pathogenesis of HCC, especially SLHCC, is crucial for developing effective intervention and therapeutic strategies to improve the outcome of patients with this devastating disease.

Recently, it has been revealed that altered expression of miRNAs contribute to the initiation and progression of cancer.23-25 Studies have shown that more than 50% of miRNAs are located in cancer-associated genomic regions or in fragile sites.2 Takata et al.26 found that miR-140 acts as a liver tumor suppressor by controlling nuclear factor kappa B (NF-κB) activity by way of directly targeting Dnmt1 mRNA. They validated that impaired miR-140 function leads to hepatocarcinogenesis,26 but its impact on HCC growth and metastasis is still unclear. In the present study, we performed a miRNA microarray to screen miRNAs relevant to HCC pathogenesis. We verified by qRT-PCR that miR-140-5p was frequently down-regulated in both HCC tissues and liver cancer cell lines. Moreover, its expression level in SLHCC was comparable to SHCC, but much higher than that in NHCC. Interestingly, miR-140-5p expression was significantly correlated with multiple nodules, vein invasion, capsular formation, differentiation, overall survival, and disease-free survival of HCC. Given that miR-140-5p was down-regulated in HCC tissues and liver cancer cell lines, we speculated that up-regulation of miR-140-5p might suppress the malignant phenotypes of HCC cells. The results derived from in vitro cell proliferation, colony formation, migration, invasion assays, and in vivo tumor formation and metastasis assays confirmed that ectopic miR-140-5p expression suppresses the potency of HCC cell proliferation and metastasis. Altogether, the suppressive effects of miR-140-5p on HCC cell growth and metastasis might contribute to a good prognosis of HCC patients with higher expression of miR-140-5p. Our findings also suggest that miR-140-5p could potentially be used as a biomarker to clinically predict metastasis, recurrence, and survival prognosis for patients with HCC.

The fundamental function of miRNAs is to regulate their target genes by direct cleavage of the mRNA and/or by inhibition of protein synthesis, according to the degree of complementarity with the target mRNA 3′-UTR.27 Multipathway reporter array is a newly technology to help us find the potential miRNA-regulated cancer signaling pathway.28 Our studies revealed that miR-140-5p suppresses the expression of TGF-β and the MAPK/ERK signaling pathway. Computational algorithms have been the major driving force in predicting miRNA targets, which are based mainly on base pairing of miRNAs and target gene 3′-UTRs.29 To explore the molecular mechanism underlying miR-140-5p function, we searched for its direct target genes using bioinformatic analysis of miRNA-mRNA 3′-UTR matching and found that TGFBR1 and FGF9 had a putative miR-140-5p binding site within their 3′-UTR. Interestingly, these two genes were related to TGF-β and the MAPK/ERK signaling pathway, respectively. Several pieces of evidence in our study also indicate that TGFBR1 and FGF9 are direct target genes of miR-140-5p in HCC. First, overexpression of miR-140-5p significantly reduced the activity of a luciferase reporter containing the 3′-UTR sequence of TGFBR1 and FGF9; second, reintroduction of TGFBR1 and FGF9 could partly abolish the effect of miR-140-5p on HCC; and third, TGFBR1 and FGF9 protein expression were posttranscriptionally down-regulated by overexpression of miR-140-5p. Pais et al.30 identified Smad3 as an miR-140 target regulated only at the protein level by using a novel methodology based on computational analysis of promoter sequences combined with mRNA microarray experiments. Then they validated Smad3 as a target of miR-140 by luciferase reporter assay and western blot assay. Moreover, they found that miR-140 suppressed the activity of the TGF-β pathway and accumulation of miR-140 was also transiently suppressed by the TGF-β pathway. These results established a double negative feedback loop for the TGF-β pathway and miR-140.30 In the present study, the expression of Smad3 protein was suppressed by miR-140-5p. Since Smad3 is a part of the TGF-β pathway and miR-140-5p suppresses the activity of the TGF-β pathway, miR-140-5p suppression of the expression of Smad3 is likely an indirect effect. TGF-β signaling is a naturally occurring potent inhibitor of cell growth.31, 32 Therefore, it is now appreciated that metastasis of most tumor types requires TGF-β activity and that, in advanced disease, TGF-β is pro-oncogenic.33, 34 This is in accordance with our study. We found that overexpression TGFBR1 could not abolish the inhibitory effect of miR-140-5p on HCC cell proliferation but suppressed HCC metastasis. On the other hand, we found that miR-140-5p suppressed HCC metastasis and HCC cell proliferation by targeting FGF9. Hendrix et al.35 identified that FGF9 possesses oncogenic activity. Abdel-Rahman et al.36 confirmed that FGF9 could activate a major intracellular effector of ERK MAP kinase. In present study, the multipathway reporter assay showed that miR-140-5p regulates the activity of ERK/MAPK signaling. Western blot analysis demonstrated that a few endogenous ERK/MAPK pathway-related proteins (such as p-ERK and H-Ras) were regulated by miR-140-5p at the protein level. Based on these results, miR-140-5p may regulate ERK/MAPK signaling through targeting FGF9. Since TGFBR1 and FGF9 both are direct targets of miR-140-5p, there might be a link between these two proteins. Yang et al.37 found that TGF-β stimulated stromal FGF-2 expression and release in vitro. Interestingly, we also found that TGFBR1 is upstream of FGF9. Combined with the results of Pais et al., who established a double negative feedback loop for the TGF-β pathway and miR-140, we would like to provide a gene regulatory network (Fig. 5G). Collectively, the down-regulation of miR-140-5p in HCC may contribute to tumor growth and metastasis, at least in part, through the up-regulation of TGFBR1 and FGF9.

In conclusion, miR-140-5p is down-regulated in HCC. miR-140-5p possesses the potency to suppress HCC growth and metastasis by regulating TGFBR1 and FGF9. Therefore, miR-140-5p could function as a tumor suppressor in HCC. The identification of miR-140-5p and its target genes, TGFBR1 and FGF9, in HCC would help in a better understanding of the molecular mechanisms underlying HCC development, which would provide us a wider perspective on HCC intervention/prevention and treatment.

Ancillary