miR‐148a inhibits early relapsed colorectal cancers and the secretion of VEGF by indirectly targeting HIF‐1α under non‐hypoxia/hypoxia conditions

Abstract Vascular endothelial growth factor (VEGF) is correlated with angiogenesis and early relapse of colorectal cancer (CRC). This study investigated the role of miR‐148a in the regulation of VEGF/angiogenesis and early relapse of CRC. We established a stable clone with miR‐148a expression in HCT116 and HT29 cell lines and created a hypoxic condition by using CoCl2 to determine the underlying mechanism of miR‐148a. The effects of miR‐148a on the phosphoryl‐ERK (pERK)/hypoxia‐inducible factor‐1α (HIF‐1α)/VEGF pathway were evaluated through Western blotting and the inhibitory effect of miR‐148a on angiogenesis was demonstrated through a tube formation assay. Sixty‐three CRC tissues (28 early relapse and 35 non‐early relapse) were analysed to assess the relationship between miR‐148a and HIF‐1α/VEGF. The protein expression of pERK/HIF‐1α/VEGF in HCT116 and HT29 cells was significantly decreased by miR‐148a (all P < 0.05). The protein expression of VEGF/HIF‐1α was strongly inversely associated with the expression of miR‐148a in the 63 CRC tissue samples (all P < 0.05). Tube formation assay demonstrated that miR‐148a significantly obliterated angiogenesis. miR‐148a suppresses VEGF through down‐regulation of the pERK/HIF‐1α/VEGF pathway and might lead to the inhibition of angiogenesis; miR‐148a down‐regulation increased the early relapse rate of CRC. This demonstrates that miR‐148a is a potential diagnostic and therapeutic target.


| INTRODUC TI ON
Colorectal cancer (CRC) is the third most common cancer worldwide and accounts for 10% of all new cancer diagnoses. Unfortunately, 20% of patients diagnosed with CRC have metastatic disease. 1 The recurrence of CRC is mostly a time-limited phenomenon; 40%-50% of recurrence events become apparent within the first year after the initial surgical resection. 2 In addition, the earlier the relapse occurs, the poorer are the overall survival rates. 3 The growth and proliferation of metastatic CRC (mCRC) depends essentially on two signalling pathways: the vascular endothelial growth factor (VEGF) and epidermal growth factor receptor (EGFR) pathways. Although substantial progress has been made in the past decades regarding the management of this disease, including surgical treatment, radiotherapy and chemotherapy, patients with advanced CRC continue to receive a poor prognosis and have a high death rate. 4 Consequently, a better understanding of the molecular mechanisms underlying CRC development and progression is urgently needed. More specifically, oligonucleotide therapies using small interfering RNAs, short hairpin RNAs, RNA aptamers and ribozymes have received considerable attention because they enable the targeted delivery of antitumour drugs without significant toxicity or other systemic side effects. [5][6][7] In this study, we focused on the use of microRNA (miRNA) as a potential therapy for CRC.
In humans, miR-148a with 68 nucleotide sequences is located on chromosome 7p15.2. miR-148a performs the common functions of many miRNA species and is implicated in a series of biological processes including cellular proliferation, apoptosis, metastasis and invasion. 8 In breast cancer cells, the ERK signalling pathway is the key downstream pathway of hypoxia-inducible factor-1α (HIF-1α) and plays an important role in angiogenesis and cancer development.
The down-regulation of miR-148a expression activates the ERK signalling pathway to increase HIF-1α and VEGF expression. 9 Moreover, hypoxia is a potential stimulator of VEGF expression and HIF-1 may regulate the hypoxic expression of VEGF in colon cancer. 10 We previously demonstrated an association between miR-148a down-regulation and early relapse in patients with CRC; this finding indicated that miR-148a is a potential biomarker for identifying high-risk patients with CRC after curative resection. 11 Tumour angiogenesis is required for tumour development and growth and HIF-1α plays a pivotal role in this process. 12 Vascular endothelial growth factor is a target gene of HIF-1α. Hypoxia-inducible factor-1α regulates VEGF expression at the transcriptional level. 13 In the present study, we identified miR-148a, which may be related to tumour angiogenesis; identified the signalling pathways that are regulated by miR-148a; and determined the role of miR-148a in the angiogenesis of CRC. Therefore, our findings provide evidence of the role and potential mechanism of miR-148a in regulating CRC angiogenesis and early relapse.

| Study design
In our previous study, we confirmed the relationship between the down-regulation of miR-148a and post-operative early relapse. 11 According to a bioinformatic analysis of pathways, 14 miR-148a could affect the function of phosphoryl-ERK [pERK]) and HIF-1α in other cancers. Hence, we suggested that that miR-148a inhibits VEGF expression by indirectly targeting HIF-1α and its relevant pathways ( Figure 1A). The design of the cell lines study is illustrated in

| Cell lines and cell lines authentication test illustration
Before the selection of cell lines, we tried to use miR-148a for transfection and proliferation in five cell lines: HCT116, HT29, SW480, SW620 and Caco-2. The results demonstrated that the HCT116 and HT29 cells showed fold changes after transfection and relative proliferation was significant altered ( Figure S1A,B) and activation of the RAS-RAF pathway has been reported to be associated with increased VEGF-induced angiogenesis. For exploring the role of miR-148a in BRAF mutation (HT29 cell) and KRAS mutation (HCT116 cell), we chose these two cell lines for experimentation.

| Patient tissue samples
For proving the expression correlation of miR-148a and target protein level, we re-analysed the expression level of miR-148a in the first cohort of 110 patients 11 and 63 CRC patients who underwent radical resection were enrolled. Among these patients, 28 were postoperative early relapse patients with miR-148a non-overexpression and 35 were post-operative non-early relapse patients with miR-148a overexpression. Early relapse was defined as local recurrence (tumor growth restricted to the anastomosis or the region of the primary operation) or distant metastasis (distant metastasis or diffuse peritoneal seeding) within one year after radical resection, and the patients who relapsed after the first year or did not relapse were placed into the non-early relapse group. 3,11,15 Written informed consent was obtained from all participants after they had been completely informed of the study protocols and that research was carried out accord-

| Construction of miR-148a overexpressing constructs
A pCDH vector (System Biosciences) was used as an miR-148a overexpression system for assessing the functional consequences of miR-148a overexpression. We constructed the pCDH-miR-148a plasmid by inserting the miR-148a polymerase chain reaction (PCR) product into the multiple cloning sites. The sequences of the primers for miR-148a were GCCTGAATTCATGCTTTTAACGAGTTATTCTTC and CTAGGCGGCCGCGCCTTGCCCCTCCCCCAAGGA. The forward primers were extended at the 5′ end to include the GAATTC sequence and the reverse primers were elongated at their 5′ end to include the GCGGCCGC sequence, which created the EcoR1 and Not1 restriction sites respectively. The constructs were confirmed through direct DNA sequencing.
F I G U R E 1 The study hypothesis and design. A, We suggested that that miR-148a inhibits vascular endothelial growth factor (VEGF) through the inactivation of the phosphoryl-ERK/ hypoxia-inducible factor-1α (pERK/HIF-1α) pathway. B, In vitro, we transfected miR-148a into HCT116 and HT29 cells and established stable colorectal cancer (CRC) clones. The protein levels of pERK and HIF-1α were examined through Western blotting and the mRNA levels of HIF-1α were tested through RT-PCR. The protein expression of VEGF was examined through ELISA miR-148a

| Establishment of a stable clone
The HCT116 and HT29 cells (5 × 10 5 ) were seeded and transfected with 400 ng of the constructs (either the negative scrambled pCDH vector or the pCDH-miR-148a plasmid) by using Lipofectamine 2000 (Thermo Fisher Scientific). To select stably transfected HCT116 and HT29 cells containing the pCDH-negative control or pCDH-miR-148a plasmid, the cells were cultured over 4 weeks in standard culture media supplemented with an additional 12 µg/mL puromycin (Sigma-Aldrich Inc, St. Louis, MO). Confirmation of stable transfection of the plasmids was obtained using a miRNA real-time quantitative PCR (RT-qPCR) assay ( Figure S1).

| miR-148a expression levels in CRC cell lines
The TaqMan

| RNA extraction and cDNA preparation
Approximately 10 7 cells were harvested from culture plates using trypsin. Total RNA, including mRNAs and miRs, was purified using Qiagen RNAeasy Columns (Qiagen, Hamburg, Germany) according to the manufacturer's protocols. For the miR assay, the cDNA of each miR was synthesized with a unique primer (Applied Biosystems) by using 20 ng of total RNA. For the mRNA quantitative assay, cDNAs were synthesized from 1 μg of total RNA with random hexamer primers by using Reverse Transcriptase (Applied Biosystems).

| mRNA expression levels
For the mRNA quantitative assay, RT-qPCR with SYBR Green

| Western blotting
Total cell lysates (20 μg) were analysed using sodium dodecyl sulphate-polyacrylamide gel electrophoresis on a 12% gel. After electroblotting onto the nitrocellulose membrane, the membranes were blocked with non-fat dry milk for 2 hours at room temperature. The membranes were then washed three times with phosphate-buffered saline (PBS) containing Tween 20 and subsequently incubated with primary antibodies (Abcam plc, Cambridge, England, UK) at 4°C overnight. Anti-HIF-1α and anti-pERK antibody were used at

| Immunohistochemistry
Formalin-fixed, paraffin-embedded blocks of CRC were collected. All 4-μm sections were dried, deparaffinized and rehydrated and heat- were positive. The staining intensity was scored as: 0, negative immunoreaction; 1, weak intensity; 2, moderate intensity; and 3, strong intensity. The sum of the two parameters varied between 0 and 6. In our study, we considered the statistical convenience and divided into low reactivity (non-overexpression), scoring ≦4 points and high reactivity (overexpression), scoring 5-6 points.

| Inhibition of HIF-1α and VEGF expression by miR-148a in the hypoxic condition
From the previous study demonstrated by Seo et al, CoCl 2 is a hypoxia mimetic agent. 18 Therefore, we used the CoCl 2 to create a hypoxic culture condition and revealed the ability of inhibition of HIF-1α and VEGF expression by miR-148a under the hypoxic culture medium.
. 19 The animals were killed 3 weeks after the tumour cells had been seeded. Tumour burdens were analysed and counted immediately without prior fixation.

| Statistical analysis
A chi-square test was used to analyse differences between the two groups (early relapse vs non-early relapse). Data are presented as the mean ± SD of three independent experiments. All statistical analyses were performed with the Statistical Package for the Social Sciences 19.0 (spss Inc, Chicago, IL). A two-tailed P < 0.05 was considered statistically significant.

| miRNA-148a inhibited the activation of pERK and HIF-1α
To investigate the effect of miRNA-148a on the activation of ERK, we demonstrated that the protein level of pERK was prominently sup-

| HIF-1α acts as an indirect target of miR-148a
To determine whether HIF-1α was a target gene of miR-148a, we compared the mRNA and protein levels of HIF-1α between cells with miR-148a overexpression and non-overexpression. The overexpression of miR-148a greatly decreased HIF-1α expression in the HCT116 and HT29 cell lines, as detected using PCR ( Figure 3A; P = 0.0026 and 0.0424 respectively) and Western blotting ( Figure 2C; P = 0.03 and 0.008 respectively), suggesting that HIF-1α is a downstream target gene of miR-148a. Furthermore, we also demonstrated that HIF-1α was not directly target gene of miR-148a using the luciferase assay in the both colon cancer cell lines ( Figure S2).

| Expression of miR-148a curbs VEGF secretion in CRC cell lines under hypoxic and no hypoxic conditions
The secretion levels of VEGF were examined through ELISA in the colon cancer cell lines, HCT116 and HT29. VEGF secretion was significantly down-regulated in the two CRC cell lines compared with in the control ( Figure 3B; P = 0.0042 and P = 0.000 76), suggesting that VEGF secretion is inhibited by the overexpression of miR-148a.

| Expression of miR-148a curbs VEGF expression in CRC tissue samples
The demographic data of the patients enrolled in this study are presented in Table 1. A strong inverse correlation was observed between miR-148a expression levels and HIF-1α and VEGF expression in CRC tissue samples, as evaluated through IHC staining (Table 2; P = 0.002 and 0.004 respectively). The overexpression of miR-148a reversed the protein expression of VEGF and HIF-1α in the tissue samples, as detected through IHC staining ( Figure 4A). These results demonstrate that overexpression of miR-148a prominently inhibits VEGF expression in vitro and in vivo.

| HUVEC tube formation assay
The HUVEC tube formation assay revealed that miR-148a inhibited angiogenesis in both CRC cell lines ( Figure 4B)  Figures S4 and S5).

| In vivo animal study: Effects of miR-148a overexpression in nude mice
To validate the role of miR-148a in tumourigenesis, we determined the effects of miR-148a overexpression on tumour growth in vivo. The

F I G U R E 2
The protein levels of phosphoryl-ERK (pERK) and hypoxiainducible factor-1α (HIF-1α) in HCT116 and HT29 cell lines were examined through Western blotting. A, The protein levels of pERK and HIF-1α were significantly decreased under overexpression of miR-148a. (Full-length blots/gels are presented in Figure S4)

| D ISCUSS I ON
A novel finding of the present study is that miR-148a can inhibit the secretion of VEGF through the indirect down-regulation of HIF-1α and its relevant pathways. Even under hypoxic condition, we also confirmed that miR-148a efficiently inhibited the expression of HIF-1α and VEGF. Regardless of in vitro or in vivo conditions, we reverified that miR-148a has the ability to inhibit angiogenesis in CRC.
Angiogenesis is a complex process through which new blood vessels are formed from an endothelial precursor. It is a critical step in cancer progression and is considered one of the hallmarks of cancer. 1 This process is mediated through a group of ligands and receptors that are tightly regulated. 20 directly. The net result of these signalling events is the increased rate of mRNA translation into HIF-1α protein. Interestingly, ERK is not only involved in regulation of HIF-1α synthesis but also its transcriptional activation. ERK phosphorylates the co-activator CBP/ p300, so it increases HIF-1α/p300 complex formation and thus stimulates its transcriptional activation function. 28 In recent years, many studies have revealed that the aberrant expression of miRNA is closely related to oncogenesis and this is now an intense field of study. miR-148a is aberrantly expressed in various cancers and has been identified as an oncogenic or tumour suppressor with crucial roles in the molecular mechanisms of oncogenesis. 8 In the some studies, it was also proven that miR148a-mediated suppression of tumour growth and tumour vascular formation existed in vivo experiments. 9,11 Overexpression of miR-148a was reported to inhibit ERBB3 expression, block downstream pathway activation (including the activation of AKT, ERK1/2 and p70S6K1) and decrease HIF-1α expression in breast cancer cell lines. 29 Some VEGF-targeted tissue and cell lines. This regulation is independent of ERK modulation by RAS/RAF pathway. Therefore, the effects of RAS/RAF mutation on ERK regulation in HCT116 and HT29 cells may be bypassed.

Number of patients
Finally, we suggested that miR-148a down-regulates VEGF through the pERK/HIF-1α pathway in CRC and might be closely associated with early relapse of CRC. However, other upstream pathways for ERK regulation may be also considered and further studies should be carried out.
The cytokine VEGF is an angiogenic factor implicated in processes such as organ development, wound healing, tissue regeneration, endothelial cell growth and vessel permeability. 34 In some solid tumours, overexpression of VEGF is associated with increased angiogenesis, growth and/or metastasis. 35,36 Researchers have also demonstrated that VEGF is not only a promising therapeutic target but also seems to be a poor prognostic factor for several cancers. [37][38][39] In CRC, VEGF-signalling-induced neovascularity is a key mediator of tumour angiogenesis, invasion and dissemination. 40 VEGF level is increased in CRC and associated with a malignancy's increased ability to spread and a poorer prognosis. 38,41 Previously, we demonstrated that VEGF played an important role in the post-operative early relapse of CRC patients, following radical resection. 27 In the current study, we further demonstrated a negative correlation between miR-148a and VEGF expression or secretion in the corresponding colorectal tissues and CRC cancer cell lines.
In conclusion, our study revealed that miR-148a down-regulated VEGF through the pERK/HIF-1α pathway. Through the inhibition of VEGF, overexpression of miR-148a might reduce post-operative early relapse in CRC patients. By using the informatics analysis and luciferase assay, we found that HIF-1α is not a potential directly target for miR-148a. Therefore, we believe that miR-148a does not directly bind to the 3′-UTR region of HIF-1α, but by inhibiting the expression of other genes. However, we would certainly take into consideration of investigating the direct target genes and the mechanism of reduced ERK phosphorylation in our future studies.

RE S E ARCH E THI C S
This study has been conducted in accordance with ethical standards and according to the Declaration of Helsinki and the national and international guidelines and has been approved by the authors' institutional review board. All processes involving the patients were approved by the Institutional Review Boards of Kaohsiung Medical University Hospital (KMUH).  Hospital (KMUH107-7R28,   KMUH107-7R29, KMUH107-7R30, KMUH107-7M22, KMUH107-7M23,   KMUH106-6M28,  KMUH106-6M29,  KMUH106-6M30,   KMUH106-6M31, KMUHS10701, KMUHS10706, KMUHS10710). In addition, this study was supported by the Grant of Biosignature in CRCs, Academia Sinica, Taiwan, R.O.C.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

AUTH O R CO NTR I B UTI O N
HL Tsai carried out the study design, molecular genetic studies, participated in the sequence alignment and drafted the manuscript. ZF Miao was responsible for molecular genetic studies and study design. YT Chen was carried out the IHC study. CW Huang and YS Yeh were responsible for data collection and analysed the data. JY Wang conceived of the study and participated in its design and coordination. All authors read and approved the final manuscript.