Background
Hepatocellular carcinoma (HCC) is the sixth most prevalent cancer and one of the leading causes of cancer-related death in both men and woman. In 2008, HCC resulted in the deaths of approximately 700,000 people worldwide, with approximately half of these deaths occurring in China [
1,
2]. Despite current knowledge and scientific advances in diagnosis and treatment modalities, the long-term survival rate of HCC still remains dismal. The unfavorable outcome could be attributed to two problems causing recurrence, intrahepatic metastasis and/or the development of de novo tumors in the remnant liver. However, the underlying biological mechanisms may differ greatly [
3‐
5]. In particular, the molecular risk factors contributing to intrahepatic metastasis and early recurrence after hepatectomy have not been well characterized.
It is well-known that cancer metastasis is a complex multistep process that involves a myriad of genetic alterations. Previous studies have identified that signaling pathways such as transforming growth factor-β (TGF-β), vascular endothelial growth factor (VEGF), Wnt/β-catenin, MAPK and small G-protein signaling play important roles in mediating HCC development [
6‐
10]. Recently, more functional players have been uncovered and are now being integrated to explain HCC initiation and progression, including microRNAs (miRNAs), long non-coding RNAs (lncRNAs) and epigenetic factors [
11‐
14].
LncRNAs are a group of poorly conserved endogenous RNA molecules longer than 200 nt in length. In the past few years, several major reports have highlighted the importance of lncRNAs in HCC metastasis, suggesting the involvement of lncRNAs in cell signaling and cancer progression. For instance, metastasis-associated lung adenocarcinoma transcript-1 (MALAT1) and lncRNA-activated by TGF-β (lncRNA-ATB) are upregulated in HCC tissues, while lncRNA low expression in tumor (lncRNA-LET) and lncRNA downregulated expression by HBx (lncRNA-Dreh) are downregulated in HCC. Furthermore, in vitro or in vivo studies have confirmed that the aberrant expression of these lncRNAs is associated with hepatoma cell proliferation or invasion [
15‐
19]. Interestingly, two lncRNAs within the Class I homeobox genes (HOX) chromosomal loci, namely, the hox transcript antisense intergenic RNA (HOTAIR) and the HOXA transcript at the distal tip (HOTTIP), were identified as being correlated with the risk of recurrence and the predicted outcomes in HCC patients [
14,
20].
In mammals, 39 HOX genes cluster on four chromosomal loci, named HOXA through HOXD, and are important for body patterning coordination during embryonic development [
21,
22]. Currently, 231 HOX-related ncRNAs have been identified including HOTAIR and HOTTIP. In this study, we report that the lncRNA HOXD cluster antisense RNA 1 (HOXD-AS1), which is expressed on the HOXD locus located on chromosome 2q31.2, plays a crucial role in HCC progression and is associated with metastasis and apoptosis phenotypes in cancer cells.
Methods
Clinical materials
Fifty cancerous and adjacent noncancerous specimens were obtained from patients with informed consent who underwent surgery for primary HCC at the Eastern Hepatobiliary Surgery Hospital (Shanghai, China) between 2010 and 2013. The study was approved by the Committees for the Ethical Review of Research involving Human Subjects from the Second Military Medical University. Among the patients, 27 had primary HCC lesions accompanied by intrahepatic metastasis at surgery (with tumor emboli in the major branches of the portal vein) and 23 had solitary HCC with no metastasis or recurrence during the two-year follow-up; the two groups were defined as the metastatic and non-metastatic groups. LncRNA and mRNA gene expression profiles were generated from six primary HCCs (three from the metastatic group and three from the non-metastatic group) and from the corresponding noncancerous hepatic fresh frozen tissues.
LncRNA microarray analysis
The LncRNA Human Gene Expression Microarray V4.0 (CapitalBio Corp, Beijing, China) was used. In brief, double-stranded cDNAs were synthesized, purified and eluted. Complementary RNA was synthesized from the eluted dsDNA products using a T7 Enzyme Mix. After amplification, the cDNA products were purified and labelled. Array hybridization was performed in a CapitalBio BioMixerTM II Hybridization Station overnight and washed. Slides were scanned and the microarray image information was converted into spot intensity values. The signal after background subtraction was exported directly into GeneSpring software for quartile normalization and further data analysis. We selected differentially expressed lncRNAs according to the following criteria: fold change >2 and P < 0.05. Hierarchical clustering analysis was employed on differentially expressed lncRNAs.
Cell culture and transfection
HCCLM3, MHCC97H, MHCC97L, SMMC7721 and L02 cells were cultured in DMEM (Biowest, Loire, France) with 10% fetal bovine serum (FBS, Biowest, Loire, France) in a humidified atmosphere containing 5% CO
2 at 37 °C. HOXD-AS1 overexpression and control pcDNA3.1 plasmids and siRNAs (GenePharma, Shanghai, China, Additional file
1: Table S1) were transfected using Lipofectamine 2000 (Invitrogen, CA, USA) according to the manufacturer’s protocols.
Total RNA was isolated from the prepared liver samples and cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Complementary DNA was synthesized following the manufacturer’s protocol (MBI Fermentas, Vilnius, Lithuania). QRT-PCR was performed with a standard SYBR-green PCR kit (TOYOBO, Osaka, Japan), and gene-specific PCR amplification was performed using the ABI 7300 (Applied Biosystems, Darmstadt, Germany). The primers are listed in Additional file
1: Table S2.
Analysis of cell motility
Cell motility was monitored using transwell and wound scratch assays. Briefly, 1 × 104 plasmid or siRNA-treated HCCLM3cells were added to the upper chamber and allowed to migrate through the polycarbonate membrane (8.0 μm PET, Millipore, Bedford, MA). After 24 h, the cells that had migrated to the lower chamber were fixed and stained with crystal violet. The wound scratch assay was performed using a pipette tip to scratch the cell layer 24 h after transfection, and phase contrast images of the wounds were recorded after 0 and 48 h.
MTT assay for cell proliferation
After transfection, cells were plated in 96-well plates at a density of 4 × 103 cells/well and incubated for the indicated times. At the end of incubation, 10 μl MTT (5 mg/mL, Sigma, USA) was added to each well, and the cells were incubated for 4 h. After staining, the samples were dissolved in DMSO, and the absorbance was recorded at 595 nm.
After treatment, cells were re-seeded in 6-well plates at a density of 500 cells/well and cultured to form nature colonies. After 10 days, the cells were washed with PBS twice and fixed with 4% paraformaldehyde for 20 min. The fixed colonies were stained by crystal purple for 10 min, photographed and counted.
Construction of stable cell lines with overexpressed HOXD-AS1 and the mouse xenograft model
To observe the effects of HOXD-AS1 overexpression on growth and metastasis in vivo, luciferase tagged cancer cells were stably infected with lentiviruses encoding HOXD-AS1 with puromycin selection. Tumor cell inoculation into the nude mice were performed as described in previously [
19]. To investigate experimental lung metastasis or liver metastasis, the anesthetized nude mouse were inoculated with different stable cell lines by tail vein injection or intra-spleen injection. Four weeks after tail vein injection or six weeks after intra-spleen injection of HCCLM3 cells, lung metastases and liver metastases were monitored by using the IVIS@ Lumina II system. Error bars show standard deviation. For tumor growth evaluation, the skin along back of mouse is incised and injected with 1 × 10
7 tumor cells. After 4 weeks the tumors in mice were removed, photographed and determined by tumor weight and tumor volumn. The animal studies were approved by the Institutional Animal Care and Use Committee of the Second Military Medical University, Shanghai, China.
Apoptosis analysis
Apoptosis was analyzed by flow cytometry using the Annexin V-PI detection kit. After transfection, cells were treated with doxorubicin (Dox, 1 μM) for 24 h and then harvested for Annexin V-PI staining according to the manufacturer’s instructions (BD Biosciences PharMingen). The double-stained cells were analyzed by flow cytometry, and the early or late apoptotic cells were measured.
Luciferase reporter assay
The 3′-UTR region of the Rho GTPase activating protein 11A (ARHGAP11A), which contains the miR-19a response element, was cloned into the pGL4.13 luciferase reporter vector to generate the luc vector. The miR-19a binding site in the luc vector was mutated to generate the luc mutant vector. To confirm the regulatory relationship between miR-19a and ARHGAP11A, miR-19a mimics, mimic NC, pcDNA3.1-HOXD-AS1, pcDNA3.1-HOXD-AS1-mut (miR-19a binding site mutation) or empty vectors were transfected into HCCLM3 cells. Forty-eight hours later, all protein extracts were analyzed using the dual luciferase reporter assay system (Promega).
Westernblot analysis
Cell samples were lysed in RIPA lysis buffer and centrifuged at 12,000 rpm at 4 °C for 15 min. Equal amounts of protein were separated on a gel and transferred to PVDF membranes (Millipore). The membranes were incubated with antibodies specific for caspase 3, caspase 9, PARP, phospho-ERK, phospho-MEK (Cell Signaling Technology, Danvers, MA, USA) and GAPDH (Epitomics, Burlingame, CA). The immunoblotting sample was incubated with horseradish peroxidase (HRP)-coupled anti-rabbit secondary antibodies (Santa Cruz, CA, USA) and visualized using enhanced chemiluminescence (Pierce, Rockford, USA).
Statistical analysis
For statistical analysis, Student’s t test was used for parametric variables; chi-square test and Fisher’s exact test (two-tailed) were used for nonparametric variables. Disease-free survival (DFS) in patients from The Cancer Genome Atlas (TCGA) dataset was analyzed using the Kaplan-Meier method and using the Gehan-Breslow-Wilcoxon test or the log-rank test for univariate analysis. All tests were performed at least three times, and a P value of less than 0.05 was considered statistically significant.
Discussion
Metastasis and the development of de novo HCC are two main causes for the poor prognosis of cancer patients [
3]. In the present study, we directly screened metastatic HCC tissue samples and non-metastatic HCC samples and identified a significantly upregulated lncRNA, HOXD-AS1, in the metastatic group (Fig.
1; Additional file
2: Table S5). HOXD-AS1 was also overexpressed in most cancerous tissue compared to the paired adjacent non-cancerous tissue (Fig.
1 and Additional file
1: Figure S1). Moreover, clinicopathological analysis revealed that overexpression of HOXD-AS1 was closely correlated with higher tumor stage and PVTT tumor invasion (Additional file
1: Table S3), indicating that HOXD-AS1 is a potential oncogene in HCC progression and metastasis. Furthermore, the pro-metastatic effect of HOXD-AS1 was demonstrated by in vitro transwell and wound scratch assays and by in vivo xenograft mouse model experiments (Figs.
1 and
2).
The underlying mechanisms by which HOXD-AS1 promoted cancer cell metastasis were then investigated. Recently, several lncRNAs were reported to act as ceRNAs by competitively binding microRNAs. For instance, linc-MD1 upregulates the expression of two transcription factors (TFs) that control muscle-specific gene expression by competitively binding to miR-133 [
28]. LncRNA-activated by TGF-β (lncRNA-ATB) “sponges” miR-200 family to regulate the expression of ZEB1 and ZEB2 and then induces EMT [
19]. Our data indicated that HOXD-AS1 could also function as a ceRNA by “sponging” miR19a, and that it could increase the expression levels of the miR19a target gene, ARHGAP11A (Fig.
4). Indeed, aberrant expression of mir19a, a possible oncogene belonging to the miR-17-92 cluster, was previously reported in multiple cancers, such as lung, colon and gastric cancers [
29‐
31]. However, a recent clinical study of 165 HCC patients revealed a differential role for miR-19a in cancer metastasis/recurrence in which miR19a was significantly downregulated in recurrent HCCs, and the miR19a downregulation was correlated with patient survival with a hazard ratio of 0.724, suggesting that miR19a is an anti-oncogene in HCC metastasis/recurrence [
25]. Consistently, our data revealed that downregulation of miR19a expression correlated well with higher tumor stage (
p = 0.0023) and PVTT tumor invasion (
p = 0.0448) (Additional file
1: Table S4). Therefore, our investigation provides a possible explanation for miR19a downregulation contributing to liver cancer progression, specifically the upregulation of the miR19a target gene, ARHGAP11A (Fig.
4). We only noticed the decrease of miR19a in a number of cancer tissues compared to the adjacent non-cancerous tissues in the metastatic group (Additional file
1: Figure S4 c-e); however, the differences in gene expression were not significant. Future studies with a larger sample size will be helpful in clarifying the biological significance of miR19a in HCC metastasis and progression.
The Ras homology (Rho) subfamily of small GTPases represents a family of small GTP-binding proteins involved in cell migration, cytoskeleton organization and proliferation. These proteins are emerging as a new class of biomarkers for cancer prognosis [
32‐
34]. ARHGAP11A belongs to the Rho GTPase activation protein (RhoGAP) subfamily, members of which promote the hydrolysis of GTP and inactivate Rho GTPases. The role of ARHGAP11A in cancer, however, is controversial. Xu and his colleagues demonstrated that ARHGAP11A accumulated in the nuclei of colorectal cancer cells, interacted with p53 and then induced apoptosis [
35]. In another study evaluating colorectal cancer development, Kagawa et al. showed that ARHGAP11A was significantly upregulated in colon cancers and that its expression levels correlated with clinical invasion status, which may result from suppression of RhoA and increased Rac1 activity [
36]. Our investigation provided evidence supporting the oncogenic and pro-metastatic effects of ARHGAP11A (Fig.
5), consistent with Kagawa’s observation.
We also examined the biological relevance of HOXD-AS1 in cancer cell growth in vitro and in vivo. This analysis revealed that HOXD-AS1 overexpression remarkably decreased Dox-induced apoptosis in HCC (Fig.
3e and f). Intriguingly, HOXD-AS1 also activates the best known G protein coupled receptor signal transduction MEK/ERK signal cascade (Fig.
6b), accompanied by decreased expression of RGS3 (Fig.
6c), which is a potential negative regulator of MEK/ERK signaling [
27]. RGS proteins serve as GTPase-activating proteins for heterotrimeric G proteins and, therefore, inactivate G protein-coupled receptor signaling pathways [
27]. Previous studies reported that several RGS proteins, including RGS2, RGS4, and RGS5, were involved in cancer development [
37‐
39]. As a member of the RGS family, RGS3 controls the signaling mediated by the Gα
q and Gα
i proteins by binding to the corresponding Gα subunits of heterotrimeric G proteins, which may cause changes in the activities of ERK, JNK and p38MAPK [
40]. Moreover, RGS3 inhibits the activation of MAPK and Akt via Gβγ subunits [
41]. Additionally, RGS3 is known to protect against cardiac hypertrophy by blocking the MEK/ERK signaling pathway [
27]. Therefore, we presumed that HOXD-AS1 may activate MEK/ERK signaling by repressing the expression levels of RGS3, which contributes to the inhibited apoptosis and accelerated proliferation during HCC progression. It has been reviewed that lncRNAs work through multiple mechanisms including ceRNAs, chromatin remodeling and natural antisense transcripts; and some lncRNAs even possess multiple mechanism characteristics, [
42,
43]. How HOXD-AS1 represses RGS3 expression in HCC cells still remain elusive. Since HOXD-AS1 can be located within nuclear or cytosolic fractions, we presumed that HOXD-AS1 may play a role in chromatin remodeling.
Acknowledgements
Not applicable.