ANGPTL1 attenuates colorectal cancer metastasis by up-regulating microRNA-138

Furthermore, the GSE29623 and GSE35982 datasets, with information on both mRNA and microRNA (miRNA), were used to identify differentially expressed miRNA between high-ANGPTL1 and low-ANGPTL1 groups. All expression profiling data in this study were downloaded from TCGA (http://​cancergenome.​nih.​gov/​) and the Gene Expression Omnibus (GEO) (http://​www.​ncbi.​nlm.​nih.​gov/​geo/​). We were able to use these databases by meeting the freedom-to-publish criteria of TCGA and NCBI.

Tumor tissue samples and paired normal mucosal tissue were obtained by surgical resection and stored at −80 °C from 2009 to 2014 at the Second Affiliated Hospital of Zhejiang University, School of Medicine. Two pathologists confirmed these tissue samples as colorectal adenocarcinoma. This project was approved by the ethical committee of the Second Affiliated Hospital of Zhejiang University, School of Medicine and informed consent was obtained from all patients.

All cells were cultured in RPMI-1640 (Gibco, Carlsbad, CA, USA) containing 10% fetal bovine serum (Life Technologies, Carlsbad, CA, USA) at 37 °C in a humidified atmosphere with 5% CO₂. All cell lines were obtained from the American Type Culture Collection (Rockville, MD, USA). The lentiviral particles containing shRNA directed against ANGPTL1 (sc-88171-V) and corresponding scramble control (sc-108080) were purchased from Santa Cruz Technologies (Santa Cruz, CA, USA), and lentivirus containing firefly luciferase was purchased from Hanbio Biotechnology (Shanghai, China). miR-138 inhibitor, mimics and negative controls were synthesized by GenePharma (Shanghai, China) and were dissolved in DEPC-treated H₂O.

The lentiviral vectors for ANGPTL1 were purchased from Cyagen Biosciences (Guangzhou, China), including pLV (Exp)-Puro-CMV > hANGPTL1/HA-IRES-eGFP and its control vector, pLV (Exp)-Puro-CMV > IRES-eGFP. One night prior to transfection, 293 T cells were plated in DMEM (Gibco) supplemented with 10% FBS without antibiotics. On the day of infection, the cells were transfected with a mixture of ANGPTL1 expression lentivector and pLV/helper packaging plasmids mix using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). The medium was replaced after overnight transfection. Supernatants were collected at 48 h post transfection, and filtered through 0.45 μm filters to remove cells and debris. Thus, the lentiviruses containing ANGPTL1/HA cDNA and the corresponding scramble control were harvested [10].

For lentiviral infection, cells were plated at 60–70% confluence. On the second day, the culture medium was replaced with complete medium containing appropriate lentiviral particles (MOI = 20) and Polybrene (2–5 μg/ml). Following 24 h of infection at 37 °C, the viral supernatant was replaced with fresh media. Another 48 h later, the infected cells were treated with 2.0 μg/ml puromycin dihydrochloride (Santa Cruz) for 2 weeks for selection of stable clones. The overexpression and knockdown efficiency was determined by quantitative real-time PCR (qPCR) and western blot (WB) analyses.

We transfected cells with a miRNA inhibitor or mimics using Lipofectamine 2000 according to the manufacturer’s instructions. Cells were seeded in 6-well plates and allowed to reach 60–70% confluence prior to transfection. The final concentration of the miR-138 inhibitor or mimics and their corresponding negative controls was 50 nmol/l. Twenty-four hours later, cells were harvested to evaluate the transfection efficiency. Then, successfully transfected cells were used for the following experiments.

For miR-138 inhibitor, the single-stranded RNA sequence was 5′-CGGCCUGAUUCACAACACCAGCU-3′. 5′- CAGUACUUUUGUGUAGUACAA-3′ was the sequence of its corresponding negative control. For miR-138 mimics, the sequences of oligonucleotides were 5′-AGCUGGUGUUGUGAAUCAGGCCG-3′ (sense), and 5′-GCCUGAUUCACAACACCAGCUUU-3′(antisense). And the sequences were 5′-UUCUCCGAACGUGUCACGUTT -3′(sense) and 5′- ACGUGACACGUUCGGAGAATT-3′ (antisense) for its negative control.

Total RNA from cells and fresh human tissues was isolated using RNAiso reagent (Takara Biotechnology, Dalian, China) according to the manufacturer’s instructions. The quality and quantity of RNA were evaluated using NanoDrop 1000 spectrophotometer (Thermo Scientific, Pittsburgh, PA, USA). cDNA was synthesized with PrimeScript™ II 1st Strand cDNA Synthesis Kit (Takara Biotechnology). To validate the mRNA expression profiles, qPCR was performed using a standard SYBR-Green PCR kit protocol (Takara Biotechnology) with the StepOne Plus Real Time PCR System (Life Technologies). The primers were synthesized by Sangon Biotech (Shanghai, China), and the sequences were as follows: ANGPTL1 forward: 5′-CAACATATTCCTAACAGCCAACAG -3′, reverse: 5′-TGACAGTCTTTGAATGGTCCTTC -3′; GAPDH forward: 5′- TCTCTGCTCCTCCTGTTCGA -3′, reverse: 5′- GCGCCCAATACGACCAAATC -3′. All PCR reactions were performed in triplicate. GAPDH was used as an internal control.

For quantifying mature miR-138, reverse transcription was performed using a miRNA 1st Strand cDNA Synthesis kit (Sangon Biotech, Shanghai, China) according to the manufacturer’s protocol. The reverse transcription primer for miR-138 was 5′-GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCGGCCT-3′, and primer for small nuclear RNA U6 was 5′- GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACAAAATA-3′. The mature miR-138 level was normalized with U6 determined by qPCR, as described previously. Primers sequences were as follows: miR-138 forward: 5′-AAGCGGAGCTGGTGTTGTGAATC-3′, reverse: 5′- ATCCAGTGCAGGGTCCGAGG-3′; U6 forward: 5′-AGAGAAGATTAG CATGGCCCCTG-3′, reverse: 5′-ATCCAGTGCAGGGTCCGAGG-3′.

WB analysis was performed as described previously [11]. Briefly, cell protein was extracted using Mammalian Protein Extraction Reagent (Thermo Scientific, Pittsburgh, PA, USA) supplemented with 1% protease inhibitor cocktails (Sigma-Aldrich, Hamburg, Germany). Protein concentration was measured using a BCA protein assay kit (Thermo Scientific). The protein samples (10–20 μg) were separated by 12% SDS-PAGE, transferred to a PVDF membrane (Bio-Rad, Hercules, CA, USA) and then detected with appropriate primary and secondary antibodies. Protein bands were visualized by chemiluminescence (Thermo Scientific) and scanned via a Kodak Image Station (Carestream Health, Inc., Rochester, New York, USA). The primary antibodies used were goat anti-ANGPTL1 polyclonal antibody (1:1000, R&D Systems, Minneapolis, MN, USA) and rabbit anti-GAPDH monoclonal antibody (1:1000, Cell Signaling Technology, Beverly, MA, USA).

Cells resuspended in 200 μl serum-free medium were seeded in the upper chamber with 10% serum-containing medium in the lower chamber of 24-well transwell plates (Corning Inc., NY, USA). After 48 or 72 h, the non-invaded cells in the upper chamber were removed with cotton swabs, and HE staining solution was then used to stain the invaded cells. Images were taken at 10× or 20× magnification. In addition, cell numbers were counted in at least 5 random microscope fields. Matrigel-coated 8 μm-pore transwells (Corning Inc.) were used for the invasion assay. The procedures and analyses were the same as those for the transwell migration assay except for the presence of Matrigel.

Cell proliferation was analyzed using the Cell Counting Kit-8 (CCK-8) (Dojindo Laboratories, Tokyo, Japan). Cells were seeded into 96-well plates at the density of 5 × 10³ cells/200 μl per well. 20 μl CCK-8 solution will be added to cells in culture and incubated for 1 h. The absorbance of each well was measured using a microculture plate reader at a wavelength of 450 nm. Three replicate wells were set up in each group and three independent experiments were performed, respectively.

Five hundred cells were inoculated into the 6-well plate containing 3 mL medium, which was changed every 3 days. After 2 weeks, clone spheres were formed. Cells were rinsed with PBS, and then fixed with 4% paraformaldehyde for 15 min. Crystal violet was added for staining for 15 min, and the plates were rinsed with flow water, and then air-dried. The number of clones in each plate was counted.

Balb/c athymic nude mice (SLAC Laboratory Animal Co. Ltd., Shanghai, China) were maintained and subjected to the experiments in accordance with the protocols approved by the Second Affiliated Hospital of Zhejiang University School of Medicine Animal Care and Use Committee. All animal experiments were performed on 5 to 6 weeks old female Balb/c athymic nude mice.

Tumor cells (1 × 10⁶) were injected subcutaneously into mice. Three weeks after injection, the mice were euthanized, and the tumors were excised. Long (L) and short (S) axes of each tumor were measured with calipers. Tumor volume (V) was calculated as follows: V = (LxS²)/2. In addition, tumor weight was also measured.

The CRC hemi-spleen liver metastasis model was established by using a previously described technique [12]. Mice were anaesthetized and an incision was made between the left abdominal and thoracic regions, and the spleen was exposed. The spleens were divided into two halves, and the halves were clipped. In total, 2 × 10⁶ cells were injected into the splenic vessels (splenic artery and veins) through one hemi-spleen followed by flushing with PBS. After the injection, the splenic vessels draining the injected hemi-spleen were clipped, and the hemi-spleen was removed. The abdominal wall and skin were sutured. Tumor metastasis was monitored using a small animal IVIS Lumina imaging system (Caliper Life Sciences, Hopkinton, MA) 15 min after intraperitoneal administration of D-luciferin (PekinElmer Inc., Boston, USA) at a concentration of 150 mg/kg. All mice were kept until death due to the neoplastic process or until the end of the experiment (2 months), and the livers were harvested for histological analysis of metastasis.

In brief, mice were anesthetized, and the cecum was exteriorized by laparotomy. A total volume of 50 μl cell suspension containing 5x10⁶ tumor cells was injected into the cecal wall by a 26G needle. Then, the injection point was slightly pressed with a cotton stick and inspected to ensure that there was no leakage. Afterward, the cecum was returned to the abdominal cavity, and the skin was closed with running sutures [13]. All mice were kept until death due to the neoplastic process or until the end of the experiment (3 months). Livers were harvested for histological analysis of metastasis.

To identify genes of tumorigenic potential, we analyzed gene expression profiles of paired cancerous and normal tissues from TCGA datasets. Gene expression analyses revealed that ANGPTL1, ANGPTL5 and ANGPTL7 were significantly down-regulated in CRC samples (Fig. 1a). Among them, ANGPTL1 was most significantly changed, and, remarkably, it was down-regulated in 87.5% (14/16) of cancer types, including breast invasive carcinoma, lung adenocarcinoma and thyroid carcinoma (Table 1). Apart from ANGPTL1, most ANGPTLs were also dramatically down-regulated in tumor tissues compared to the corresponding normal samples (Additional file 1: Table S1). Next, we validated our finding in the GSE32323 cohort from the GEO database. Analysis of 17 paired CRC and normal samples showed that ANGPTL1 was down-regulated in tumor samples (P = 0.01) (Fig. 1b). Finally, we confirmed these results in our oncology center. The basic characteristics of these patients are specified in Additional file 2: Table S2. Unsurprisingly, the results were similar to what we had found in TCGA and GEO datasets (P = 0.02, Fig. 1c). The low expression of ANGPTL1 in CRC and many other cancer types suggested that it was a potential suppressor of cancer and may be associated with CRC initiation and progression.

To further determine the clinical significance of ANGPTL1 expression, we explored its association with survival information in TCGA dataset. The results showed that patients with high ANGPTL1 mRNA level had significantly longer OS than patients with low level of ANGPTL1 in stage IV metastatic CRC (P = 0.03, Fig. 1d). Moreover, in the GSE24550 cohort, high expression of ANGPTL1 predicted better disease-free survival in stage III patients (P = 0.03, Fig. 1e). Taken together, these results suggested that high ANGPTL1 expression was correlated with better survival in CRC patients.

To investigate the little-known function of ANGPTL1 in the biological behavior of CRC cells, ANGPTL1-targeting shRNA or ANGPTL1 cDNA plasmids packaged in lentiviruses were used to modulate ANGPTL1 gene expression in human CRC cell lines SW480 and SW620. Overexpression and knockdown efficiency of ANGPTL1 was confirmed by qPCR and WB (Additional file 3: Figure S1A). At first, we assessed the effect of ANGPTL1 on tumor cell proliferation using the CCK-8 kit, and the results showed no significant difference between SW620-ANGPTL1 and SW620-Ctrl cells (Additional file 3: Figure S1B, P = 0.10). No significant difference was found by colony-forming assay as well (Additional file 3: Figure S1C, P = 0.27). Similar results were found in SW480-shANGPTL1 and SW480-Ctrl cells (Additional file 3: Figure S1D, E).

To further confirm our findings, we compared the growth of SW620-ANGPTL1 and SW620-Ctrl cells in a subcutaneously inoculated model. Cells were injected subcutaneously into the flanks of nude mice (N = 5 for each group). Three weeks after injection, tumor weight (P = 0.29) and volume (P = 0.53) showed no significant differences between SW620-ANGPTL1 and SW620-Ctrl group (Additional file 3: Figure S1F-H). Taken together, these results indicated that ANGPTL1 does not significantly affect proliferation and colony formation of CRC cells.

Next, we examined the role of ANGPTL1 in migration and invasion by transwell assay. As shown in Fig. 2a, SW620-ANGPTL1 cells exhibited remarkably reduced mobility compared to vector control cells (P < 0.0001). Matrigel-coated invasion assay also indicated that SW620-ANGPTL1 cells displayed significantly decreased invasion capacity compared to SW620-Ctrl cells (P < 0.0001) (Fig. 2b). By contrast, the number of invading cells was significantly increased in SW480-shANGPTL1 cells compared to that in SW480-Ctrl cells in both transwell migration (P < 0.0001, Fig. 2c) and invasion assay (P = 0.02, Fig. 2d). In addition, knockdown of ANGPTL1 in LoVo cells produced the same results as in the SW480 cells (Additional file 4: Figure S2). It can be inferred that the decrease in the number of invading cells by transwell assay is due to the effect of ANGPTL1 on migration and invasion rather than on proliferation, because ANGPTL1 has no significant effects on cell proliferation. Collectively, these results indicated that ANGPTL1 represses migration and invasion of CRC cells in vitro.

The liver is the most common site of metastasis in CRC. To investigate the effect of ANGPTL1 on the metastatic potential of CRC cells, we employed a hemi-spleen liver metastasis model, in which tumor cells suspensions were injected into a part of the spleen, with successive hemi-splenectomy. The other half of the spleen, uncontaminated with tumor cells, was retained in the animal. As the tumor cells were inoculated into the portal vein, diffuse liver metastasis was established [14, 15]. Additionally, the orthotopic injection model, with cancer cells injected into the cecum for local tumor growth, better mimics the natural trajectory of colon cancer in the patients [16]. Therefore, this model was also used to further evaluate spontaneous formation of liver metastasis.

In the hemi-spleen liver metastasis model, mice were sacrificed 2 months after injection of tumor cells, and livers were harvested for histological examination of metastasis (Fig. 3a). We found that the incidence of liver metastasis was lower in the SW620-ANGPTL1 group compared to that in the SW620-Ctrl group (14.29% vs 28.57%) (Table 2). Moreover, we used the IVIS Lumina imaging system to monitor the migration of tumor cells in vivo on day 14. The fluorescence intensity of liver lesions was higher in the SW620-Ctrl group (Fig. 3b). Apart from liver metastasis, we found that the rate of extrahepatic metastasis in SW620-ANGPTL1 cells was also lower than that in SW620-Ctrl cells using the in vivo imaging system (Fig. 3b). In addition, the mice in this group displayed longer OS (P = 0.02, Fig. 3c). Similarly, in the orthotopic injection model, the rate of liver metastasis was decreased in the SW620-ANGPTL1 group (14.29% vs, 28.57%) (Table 2), which was confirmed by histological analysis.

By contrast, the incidence of liver metastasis in the SW480-shANGPTL1 group increased to 21.05%, whereas the control group had a very low rate of liver metastasis (0.06%) in the hemi-spleen liver metastasis model (Table 2). Additionally, in the orthotopic injection model, the rate of liver metastasis was higher in the SW480-shANGPTL1 group compared to that in the control (30.00% vs 11.11%). Collectively, high ANGPTL1 expression level is associated with a reduction in CRC metastasis and better prognosis in mice with CRC.

Accumulating evidence has demonstrated critical roles for miRNAs in various cellular processes, including tumor development and progression. Kuo et al. [7] demonstrated that ANGPTL1 regulates the expression of miR-630 at the transcriptional level, which targets SLUG and results in the inhibition of lung cancer cell metastasis. Thus, we considered whether miRNAs are involved in the ANGPTL1-mediated inhibition of migration of CRC cells, as well as the specific miRNAs contributing to this process. To further verify this hypothesis, we extracted the profiling data for both miRNAs and mRNAs from the GSE29623 dataset to address the relationship between ANGPTL1 and miRNAs in CRC. The 8 samples with the lowest ANGPTL1 mRNA level were compared with the 8 samples with the highest ANGPTL1 mRNA level, and a heatmap was formulated to measure the diversity of miRNA level (Fig. 4a). Among these candidates, miR-138 was reported to be down-regulated in CRC tissues, and its down-regulation was associated with more severe metastasis in vitro and in vivo by targeting TWIST2 [17]. miR-138 was suggested to be of great priority of being involved in ANGPTL1-regulated metastasis. To validate this finding, we explored the association between transcript levels of ANGPTL1 and miR-138 in another GSE dataset (GSE35982). Pearson correlation analysis suggested that ANGPTL1 mRNA level was positively correlated with miR-138 level in CRC tumor samples (Pearson correlation value = 0.94, P = 0.001). However, no significant linear relationship was found in normal samples (Pearson correlation value = -0.45, P = 0.27). The specific data for expression levels of ANGPTL1 and miR-138 in tumor tissues are plotted in Additional file 5: Figure S3A.

The level of miR-138 was further determined in cells with ANGPTL1 overexpression and knockdown. As shown in Additional file 5: Figure S3B, overexpression of ANGPTL1 significantly enhanced miR-138 expression by 1.51-fold (P < 0.0001), whereas miR-138 level was markedly (86%) inhibited by knockdown of ANGPTL1 (P < 0.0001), indicating that ANGPTL1 may regulate the expression of miR-138. Then, we conducted transwell migration assay to determine whether miR-138 is involved in ANGPTL1-mediated inhibition of migration of CRC cells. Cells were transiently transfected with a miR-138 inhibitor, mimics or negative controls at 50 nmol/l using Lipofectamine 2000. As shown in Fig. 5a-b, SW620-Ctrl cells treated with miR-138 inhibitor showed enhanced migratory capacity (P = 0.01), and miR-138 inhibitor reversed the inhibition of migration in SW620-ANGPTL1 cells (P = 0.001). In addition, in SW480-shANGPTL1 cells transfected with miR-138 mimics, the shANGPTL1-induced increase in migratory capacity was significantly attenuated (P = 0.006, Fig. 5c-d). These findings suggest that ANGPTL1 directly or indirectly up-regulates the expression of miR-138, and miR-138 is involved in ANGPTL1-mediated inhibition of migration of CRC cells.