Down-regulation of Krüppel-like factor-4 by microRNA-135a-5p promotes proliferation and metastasis in hepatocellular carcinoma by transforming growth factor-β1

Krüppel-like Factor-4 (KLF4) is a zinc finger transcription factor which plays an important role in cell cycle, proliferation and apoptosis. In Hepatocellular Carcinoma (HCC), the function of KLF4 has been characterized as tumor suppressor. However, the mechanism remains largely unknown. In this study, we demonstrated that TGF-β1 down-regulated KLF4 by activating miR-135a-5p. MiR-135a-5p promoted proliferation and metastasis in HCC cells by direct targeting KLF4 both in vitro and in vivo. In addition, miR-135a-5p expression was up-regulated in clinical HCC tissues, and was inversely correlated with the expression of KLF4. Taken together, our data indicated that TGF-β1 down-regulated KLF4 by activating miR-135a-5p, promoting proliferation and metastasis in HCC.


INTRODUCTION
Hepatocellular Carcinoma (HCC) is the most common primary liver cancer in adults. The incidence rate of liver cancer fell from fifth to sixth place while the mortality rate rose from third to second in the world and more than 50% of patients who are diagnosed with liver cancer are reported to be Chinese [1]. Recent studies have demonstrated that hepatocarcinogenesis linked with two main pathogenic mechanisms: (1) cirrhosis associated with hepatitis infection (for example, HBV or HCV), toxins (for example, alcohol or aflatoxin) or metabolic influences, and (2) mutations of oncogenes or tumor suppressor genes [2]. The mechanisms are involved in several important cellular signaling pathways, including EGF/EGFR [3], PI3K/AKT/mTOR [4,5], RAF/MEK/ERK [6], HGF/c-MET [7] and WNT/β-catenin [8]. Besides that, clinical researches suggest that molecularly targeted drugs block critical signaling pathways to treat for advanced HCC. For example, sorafenib (a small-molecule kinase inhibitor) could prolong advanced HCC patients' overall survival (OS) for around 2-3 months [9]. Therefore, it is beneficial for the treatment of HCC to find a new target.
Krüppel-like Factor-4 (KLF4) is a zinc finger transcription factor which plays an important role in cell cycle, proliferation and apoptosis [10][11][12][13]. KLF4 has been confirmed as both tumor suppressor and tumor promoter in different types of cancer [14,15]. In HCC, the function of KLF4 has been characterized as tumor suppressor including: (1) KLF4/VDR signaling pathway may prevent the progress of HCC [10], (2) overexpression of KLF4 inhibits tumor proliferation, invasion and migration in HCC cells and reverts EMT [16], and (3) high expression level of KLF4 is associated with better survival [17]. Additionally, KLF4 activates TGF-β signaling by binding to the TCE of the AT1R promoter in vascular smooth muscle cells [18]. Decreasing KLF4 and increasing SLUG expression is involved in EMT signaling in advanced primary prostate cancer [19].
In this study, we used a miRNA library to identify that miR-135a-5p is a regulator of KLF4, and show their potent effects on TGF-β1 signaling pathway in HCC. Moreover, we found that miR-135a-5p expression was increased in clinical HCC tissues, with concomitant low levels of KLF4, suggesting that up-regulation of miR-135a-5p may be involved in hepatocarcinogenesis.

Down-regulation of KLF4 by TGF-β1
Previous studies have indicated that TGF-β induced EMT by both Smad-dependent and Smad-independent pathways which acquire the capacity to detach and migrate away from the primary tumor [23,24]. Moreover, loss of E-cadherin is considered as a landmark event of EMT that initiates a series of signaling events and major cytoskeletal reorganization [23]. Interestingly, KLF4 is also considered to be associated with EMT, which the expression is reduced during EMT process [19,25]. Thus, we speculated that TGF-β might regulate the expression of KLF4. To examine this possibility, Bel-7402 was treated with TGF-β1 for various concentrations (0-12.5ng/ml). Activation of the EMT signaling pathway was confirmed by E-cadherin [23]. As expected, the protein level of KLF4 was reduced using TGF-β1 treatment for 24h. Moreover, the higher concentration of TGF-β1 treatment, the more obvious decrease of KLF4 level ( Figure 1A). KLF4 mRNA was also decreased after TGF-β1 treatment ( Figure 1B). Unlike KLF4, no changes in GAPDH protein or β-actin mRNA levels were observed using TGF-β1 treatment, which justified the use of GAPDH and β-actin as a control. Similar results were observed in other three cells SK-Hep-1, Hep-3B and Huh 7 ( Figure 1C and Figure  1D). All these results demonstrated that TGF-β1 repressed the expression of KLF4 in both protein and mRNA levels in various HCC cells.
To examine whether the EMT signaling pathway was activated by TGF-β1, we detected the expression of E-cadherin by qRT-PCR. The results showed that E-cadherin mRNA level was down-regulated more than 60% after TGF-β1 treatment ( Figure 1E).

Up-regulation of MicroRNA-135a-5p by TGF-β1
MicroRNAs are known to be involved in various biological-signaling pathways. The reduction of KLF4 expression in response to TGF-β1 treatment suggested that a miRNA could be involved. Thus, we selected 11 miRNAs including miR-206, miR-26b-5p, miR-214-3p, miR-375, miR-449a, miR-135a-5p, miR-9-5p, miR-107, miR-128-1-5p, miR-363-3p, miR-367, which predicted to target KLF4 by binding to regions in the 3′-UTR (Prediction websites: Target Scan Human and MicroRNA. org). So we purchased the 11 miRNAs mimic (Rib Bio, China) as a library to analyze the expression of KLF4 after transfecting them in Hep-3B, the results showed that the expression of KLF4 was almost decreased (Figure 2A). Furthermore, we also tested the expression of the 11 miRNAs after TGF-β1 treatment and only miR-135a-5p was induced at least 1.5-fold ( Figure 2B), suggesting that it might be critical for TGF-β1 to mediate reduction of KLF4. To verify the data, we used qRT-PCR to follow the expression of mature miR-135a-5p in other three cells using TGF-β1 treatment, and results were similar ( Figure 2C).
Next, we explored the mechanism of miR-135a-5p induction by TGF-β1. miRNAs can be regulated at the level of transcription of the pri-miRNA, or at the Drosha or Dicer processing steps [26].
So we examined whether the expression level of pri-miR-135a-5p is regulated by TGF-β1 in HCC cells. After TGF-β1 treatment for 24h, the level of pri-miR-135a-5p was elevated relative to untreated cells ( Figure 2D). Therefore, the results indicated that TGF-β1 is able to enhance the transcription of miR-135a-5p.

MicroR-135a-5p is critical for down-regulation of KLF4 by TGF-β1
To examine whether miR-135a-5p is able to downregulate KLF4 in HCC cells, miR-135a-5p mimic and miR-135a-5p inhibitor were transfected into HCC cells, respectively. KLF4 protein and mRNA levels were decreased in miR-135a-5p mimic, and increased in miR-135a-5p inhibitor ( Figure 3A and Figure 3B). To further confirm the direct association of miR-135a-5p with the 3′-UTR of KLF4, we predicted the miR-135a-5p binding site in the human KLF4 3′-UTR by TargetScan 6.2, and the 3′-UTR of KLF4 mRNA containing the wild type or mutated putative miR-135a-5p binding sequence was cloned into a luciferase reporter vector, respectively ( Figure 3C, left). The luciferase reporter assay demonstrated that miR-135a-5p could effectively inhibit the luciferase activity in the wild type and completely abrogate its regulatory activity in mutation ( Figure 3C, right). These results indicated that the level of KLF4 expression was negatively regulated by miR-135a-5p in HCC cells.
Next, we detected the expression of E-cadherin in HCC cells, which were transfected miR-135a-5p mimic and miR-135a-5p inhibitor, to examine whether the EMT signaling pathway was activated by miR-135a-5p. The results showed that E-cadherin mRNA level was decreased after transfecting miR-135a-5p mimic, and increased after transfecting miR-135a-5p inhibitor ( Figure 3E).

The relationship between MicroR-135a-5p and KLF4 in HCC patients
To evaluate the clinical relevance of miR-135a-5p induced down-regulation of KLF4, we analyzed the expression of miR-135a-5p and KLF4 in HCC specimens by qRT-PCR. Patient characteristics and clinical features were observed in Table 1. The results showed that the levels of miR-135a-5p were significantly increased in 27 of 44 clinical HCC tissues relative to the adjacent nontumorous tissues ( Figure 5A). Using linear regression analysis, we found that there was a significant negative correlation between miR-135a-5p and KLF4 expression in HCC tissues ( Figure 5B). TGF-β1 and KLF4 had the similar negative correlation ( Figure 5C). These results suggested that the relationship between TGF-β1, miR-135a-5p and KLF4 in HCC tissues was consistent with our phenotypic assays in HCC cells.

DISCUSSION
In this study, we demonstrated a critical role and mechanism of TGF-β1 down-regulating KLF4 by activating miR-135a-5p. Moreover, miR-135a-5p promoted proliferation and metastasis as a novel oncogenic miRNA in HCC. In advanced cancer, TGF-β was often overexpressed, and functioned as a promoter by stimulating EMT which strengthens invasiveness and metastasis [27,28]. TGF-β signaling toward EMT was mediated by both Smad-dependent and Smad-independent pathways, and the Smad pathway was unique to TGF-β signaling [29][30][31]. Keratinocyte-specific Smad2 ablation resulted in increased epithelial-mesenchymal transition during skin cancer formation and progression [32]. Smad3 ablation in the liver prevented hepatocytic EMT showing that Smad3 was essential for EMT [33,34]. In addition, E-cadherin expression was induced during TGF-β1-mediated EMT in breast cancer [29]. In this study we demonstrated that E-cadherin expression was decreased after TGF-β1 treatment ( Figure 1E), which meant the EMT signaling pathway was exactly activated by TGF-β1.
Previous studies on EMT showed that some transcription factors were controlled by miRNAs. The miR-200 family and miR-205 regulated epithelial to mesenchymal transition by targeting the E-cadherin repressors ZEB1 and ZEB2 [35][36][37][38]. Also, TGF-β down-  regulated the miR-200 family by inhibiting Akt1 which increased the abundance of ZEB1 and ZEB2 [39]. In addition, the induction of a contractile phenotype in human vascular smooth muscle cells by TGF-β and BMPs was mediated by miR-21 [40]. In this study we developed a miRNA mimic library to identify miR-135a-5p functions, we found that both mature miR-135a-5p and pri-miR-135a-5p levels were increased after TGF-β1 treatment ( Figure 2C and Figure 2D), indicating that TGF-β1 could enhance the transcription of miR-135a-5p. Moreover, we detected the expression of E-cadherin after transfected miR-135a-5p mimic and miR-135a-5p inhibitor, and the results showed that the EMT signaling pathway was also activated by miR-135a-5p ( Figure 3E). Thus, we speculated that the promoter region of miR-135a-5p might have a direct binding to the related genes in TGF-β signaling pathway, and Smad1/4 or Smad2/3 were the most potential. This speculation needs further validation in the future. KLF4 was a complex transcription factor that can act as a transcriptional activator, a transcriptional repressor, an oncogene, and a tumor suppressor [41]. It was reported that KLF4 negatively regulated EMT of Gastrointestinal Cancers through crosstalk with TGF-β, Notch, and Wnt signaling pathways [25]. In breast cancer, KLF4 inhibits epithelialto-mesenchymal transition through regulation of E-cadherin gene expression [42]. In this study, we demonstrated the expression of KLF4 after TGF-β1 treatment, the results showed that KLF4 was down-regulated by TGF-β1 at both protein and mRNA level ( Figure 1C and Figure  1D). Furthermore, researches prove that miR-7 promotes epithelial cell transformation by targeting KLF4 [43]. MiR-29a promoted colorectal cancer metastasis by regulating matrix metalloproteinase 2 and E-cadherin via KLF4 [44]. Consistent with this idea, in this study, we showed that miR-135a-5p enhanced proliferation and metastasis in HCC by targeting KLF4 both in vitro and in vivo (Figure 4). We also demonstrated that transfected miR-135a-5p mimic in HCC cells, both protein and mRNA levels of KLF4 were significantly decreased ( Figure 3A and Figure 3B). Importantly, our results from the luciferase reporter assays showed that miR-135a-5p directly interacted with the KLF4 3′-UTR ( Figure 3C). And the relationship among TGF-β1, miR-135a-5p and KLF4 was shown in Figure 3D. Therefore, our findings were consistent in supporting the functional significance of TGF-β1 down-regulating KLF4 by activating miR-135a-5p in hepatocellular carcinoma ( Figure 6). Figure 6: Schematic of TGF-β1 down-regulating KLF4 by activating miR-135a-5p. Upon TGF-β1 stimulation, pri-miR-135a-5p expression is able to induce through Smad2/3 or Smad1/4 which is processed into mature miR-135a-5p. Then mature miR-135a-5p represses KLF4 expression by binding to the 3′-UTR of KLF4. And the decrease of KLF4 may further inhibit the expression of VDR. TGF-β1, miR-135a-5p and KLF4 have an impact on the EMT signaling pathway. www.impactjournals.com/oncotarget In summary, this study revealed a mechanism of TGF-β1 down-regulating KLF4 by activating miR-135a-5p in HCC. Moreover, miR-135a-5p promoted proliferation and metastasis as a novel "oncomiR" by down-regulating KLF4 whereby directly interacts with KLF4 3′-UTR in HCC. These data suggested that miR-135a-5p and KLF4 should be further explored as potential diagnostic and prognostic markers in hepatocellular carcinoma. We anticipate that this defined loop pathway will provide insight for investigating the regulation of EMT that is proposed to function during tumor progression.

Cell lines and culture conditions
Four human HCC cell lines SK-Hep-1 (a liver cancer cell), Hep-3B, Bel-7402, Huh 7 were purchased from Shanghai Cell Bank of Chinese Academy of Sciences (Shanghai, China) and cultured in Dulbecco's modified Eagle medium (DMEM, Invitrogen, USA) supplemented with 10% fetal bovine serum (FBS, Hyclone) as well as 100U/ml penicillin and 100μg/ml streptomycin. All of the cells were maintained in a humidified incubator at 37°C with 5% CO2. For TGF-β1 stimulation experiments, 10ng/ ml of TGF-β1(R&D systems,USA) was used.

Western blotting
The HCC cells were lysed and extracted into protein with 1× SDS-PAGE loading buffer. Protein concentrations were determined using the BCA protein assay kit (Beyotime Institute of Biotechnology, China). Equal amounts of protein were separated on 10% SDS-PAGE gel and then transferred to the PVDF membranes. After blocking with 5% skim milk for 2 hours, the membranes incubated with primary antibodies as follows: KLF4 www.impactjournals.com/oncotarget (1:500, Santa Cruz Biotechnology) and GAPDH (1:3000, Santa Cruz Biotechnology).

CCK-8 assay
The Cell Counting Kit-8 (CCK-8) assay kit (Dojindo, Japan) was used to test the effect of miR-135a-5p on cell proliferation. Transfected cells were plated into 96-well plates at a density of 2×103 cells per well; 10 μl CCK-8 solution was added to each well the next day, totally six days. The cells were incubated for 70 minutes and the absorbance at 450nm was tested using enzymelinked immunosorbent assay reader (Dasit, Milan, Italy). Each experiment was conducted for at least three times and the average of the results was analyzed.

Cell migration assay
The transfected HCC cells in serum-free media were plated into the upper part of a transwell chamber in a 24-well format with 8 mm diameters (Corning, USA) at a suitable density. In the bottom chamber, 800μL of normal DMEM medium containing 10% FBS was added as a chemoattractant and the chambers were incubated for 24-48h at 37°C with 5% CO2. The cells on the upper part were removed by cotton swap and the migrated cells were stained with 0.05% crystal violet for 30 minutes. At last, counting the migrated cells under a microscope in five random fields and each experiment was conducted for at least three times and the average of the results was analyzed.

Wound healing assay
The transfected HCC cells in DMEM medium containing 10% FBS were plated into six-well plates at a density of 90%. Wounds were created by a plastic tip, cell debris were removed using PBS, and replaced by 0.5% FBS-containing DMEM. The scratched cells were incubated at 37°C with 5% CO2. The initial and residual scratched gap breadth were measured by the light microscope (Nikon, Japan) at 0h, 24h and 48h, respectively. Each experiment was conducted for at least three times.

Animal experiments
Male athymic nude mice (4 weeks old) were purchased from Animal Center of the Chinese Academy of Science (Shanghai, China) and maintained in sterile laminar flow cabinets. Mice were randomly divided into four groups and inoculated subcutaneously with 1×10 6 cells in 100μl PBS at the age of 6 weeks old, respectively. Similarly, mice were tail vein injected with 1×10 6 cells in 100μl PBS at the same time. After 6 weeks, animals were sacrificed, the tumors and lungs were removed and weighed. The lungs were fixed in formalin overnight before evaluating lung metastasis, and the numbers of metastatic nodules on the lung surface were counted. All animal care and procedures were approved by Tongji University School of Shanghai East Hospital for Animal Experiments.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 6.0. Differences between two groups were explored by Student's t test. For comparison of paired tissues, the values were presented as mean ± SEM. Only a P value of less than 0.05 was considered significant. "*" indicates P<0.05; "**" indicates P<0.01; "***" indicates P<0.001.