Introduction
Hepatocellular carcinoma (HCC) is the most frequent hepatic malignancy and has a poor prognosis [
1]. To date, the only option for curative treatment for HCC is surgical resection or liver transplantation, but most patients are usually diagnosed at an advanced stage and are not candidates for surgery. Moreover, tumors commonly reappear after surgery, and the 5-year postoperative survival rate is poor [
2]. Therefore, a deeper investigation of the mechanisms associated with HCC progression and recurrence is of clinical significance and can lead to the development of new therapeutic approaches for patients with HCC [
3].
The discovery of the constitutive activation of the mitogen-activated protein kinase (MAPK) pathway as a result of gene mutation or amplification in cancer has revolutionized traditional cancer medicine by enabling precision medicine for a subgroup of patients [
4]. Furthermore, the emergence of cancer immunosurveillance introduced immune checkpoint blockades to cancer therapy [
5]. However, the majority of patients treated with these agents show a minimal clinical response and sometimes develop drug resistance, which leads to further disease development and death [
6]. Thus, the molecular mechanisms of HCC are still being studied.
Many molecular events have been linked to the development and progression of HCC, such as overactivation of
c-Myc and inactivation of
Pten and
p53 [
7]. Over the past few decades, significant advances have revealed chromosome aberrations in HCC, including gains of 1q21–23 and 8q22–24, that are involved in the early development of HCC, and the gain of 3q22–24 has been associated with tumor recurrence and poor survival [
8]. Chromosomal amplification at 7q21-7q31 was reported to be closely related to tumor recurrence [
9‐
12], and several oncogenes, such as
MET and
PEG10, in this region are amplified in many cancers [
13,
14]. Thus, specific chromosomal aberrations can be responsible for carcinogenesis via dysregulation of tumor suppressor genes or oncogenes. Considering the hundreds of genes in a chromosomal region, more work is needed to reveal additional tumor-related genes in regions with chromosomal aberrations [
15].
Circular RNAs (circRNAs), a novel kind of regulatory RNA characterized by a continuous covalent closed loop without a 5′-cap structure or 3′-poly A tail, are considered splicing error byproducts. Hundreds of circRNAs exist in mammalian cells and regulate a broad range of biological processes through various mechanisms, including sponging of microRNAs (miRNAs) [
16]. Aberrant expression of circRNA has also been implicated in the initiation and development of various diseases, including cancers [
17]. For example, circTRIM33–12 acts as a sponge of miRNA-191 to inhibit HCC progression [
18], while circHIPK3 is a tumor suppressor in bladder cancer [
19]. As circRNAs are novel endogenous noncoding RNAs, the biological functions of most circRNAs and their underlying mechanisms in the pathogenesis of HCC remain largely unclear and need further exploration [
20].
Most cells cannot survive an attack from the immune system, with the exception of tumor cells due to their loss of immunogenicity and immunosuppression at malignant mass sites [
21]. Evading immune destruction has been deemed an emerging hallmark of cancer, and the crosstalk between the immune system and tumor cells plays a definitive role in progression to advanced stages of cancer [
22]. Moreover, the host immune system can combat cancers and improve outcomes for cancer patients, which has been documented by clinical testing of immune checkpoint blockades [
23]. A programmed cell death 1 (PD1) antibody has been approved for second-line therapy in advanced HCC [
24]. However, only 17–18% of patients with advanced HCC have achieved complete or partial response to anti-PD1 antibody therapy, and adverse events have been observed, indicating that a better understanding of cancer immunosuppression is needed [
25].
Previous work has implicated chr.7q21-7q31 amplification in the progression of HCC [
9,
26,
27]. This region contains critical oncogenes involved in tumor progression, such as MET proto-oncogene receptor tyrosine kinase (
MET)
, hepatocyte growth factor (
HGF)
, SEM1 26S proteasome complex subunit (
SHFM1)
, minichromosome maintenance complex component 7 (
MCM7), and ATP binding cassette subfamily B member 4 (
ABCB4, 12]. Therefore, we speculated that chr.7q21-7q31-derived circRNAs could act as oncogenes or tumor suppressors in HCC. Here, we determined the expression of circRNAs in the 7q21-7q31 region in human HCC tissues and correlated this expression with HCC patient survival and disease recurrence. We modified the expression of the identified circRNAs to determine whether they can promote HCC development in HCC cells in vitro. Modified cells were also implanted into immunocompetent mice to assess the effects on tumor development. Then, we performed additional experiments to determine the mechanism of action of these effects.
Methods
Cell lines, animals and transfection of lentiviral vectors
We used the human HCC cell lines HCCLM3 (Liver Cancer Institute, Zhongshan Hospital), and Huh7, Hep3B, and HepG2 (American Type Culture Collection) with high/low metastatic capacity and the murine HCC cell line Hep1–6 (American Type Culture Collection). Cells were grown at 37 °C and 5% CO2.
C57BL/6 mice were cultured in specific pathogen-free conditions (Shanghai Institute of Material Medicine, Chinese Academy of Science). Animal care and experimental protocols were in line with the guidelines of the Shanghai Medical Experimental Animal Care Commission.
Lentiviral vectors containing circMET, miR-30a-5p, snail, shcircMET, shmiR-30a/e-5p, simiR-30b-5p, simiR-30c-5p, or simiR-30d-5p were constructed (Shanghai Genomeditech Co. Ltd., Shanghai, China). Stable transfectants were characterized by quantitative real-time polymerase chain reaction (qRT-PCR) or western blotting. The targets of sh/si-miR-30-5p are listed in Additional file
1: Table S1.
Patients and follow-up
Specimens obtained from the vicinity of the tumor margin were collected from 209 patients with HCC who underwent radical resection at the Fudan University Liver Cancer Institute (Shanghai, China) from 2006 to 2008. Ethical approval was confirmed by the Zhongshan Hospital Research Ethics Committee, and written informed consent was obtained from each patient. Follow-up data were ended by March 2014, and the follow-up median time was 62 months (range 4 ~ 121 months).
circRNA immunoprecipitation (circRIP) and in situ hybridization
circRNAs precipitation in vivo, circRNAs immunoprecipitation and in situ hybridization were performed as previously described with minor modifications, and the procedures are described in the Additional file
4: supplementary materials and methods [
18,
28].
Quantitative real-time polymerase chain reaction analysis, western blotting analysis, immunofluorescence assays, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium dromide assay, cell migration, Matrigel invasion assays, and gene microarray
These experiments were performed according to previous reports [
14,
18,
29], and the procedures are described in the Additional file
4: supplementary materials and methods.
Human CD8+ T cell isolation and preparation and CD8 T cell chemotaxis assay
Blood was collected from HCC patients to isolate peripheral blood mononuclear cells using Ficoll. Then, CD8
+ T cells were purified using anti-CD8 antibody-coupled immunomagnetic beads and resuspended in RPMI 1640 in the supplemented with 2% FBS. CD8
+ T lymphocyte chemotaxis was assayed in a Transwell system (Corning, Tewksbury, MA, USA) using 5-μm polycarbonate membranes, as described previously with modifications [
30]. Human CXCL10 (50 ng/ml) was added to RPMI 1640 supplemented with 2% FBS and incubated with HCC cells for 24 h, after which the supernatant was added to the bottom wells. Then, CD8
+ T lymphocytes from the peripheral blood of HCC patients suspended in RPMI 1640 with 2% FBS were added to the top wells and incubated 2 h. Following incubation, CD8
+ T lymphocytes that migrated to the lower chamber were counted using a cell counter. The chemotactic index was evaluated as the ratio of the number of CD8
+ T lymphocytes that migrated to CXCL10-containing supernatant wells divided by the number of CD8
+ T lymphocytes that migrated to RPMI 1640 alone.
Immunohistochemistry (IHC)
Immunohistochemical staining of the target circRNAs and proteins was carried out on a tissue microarray (TMA). The circMET expression was quantified as follows: three images of representative fields of circMET staining were captured under a computerized image system. Images were analyzed with Image-Pro Plus version 6.2 software (Media Cybernetics) using a special function called measurement of integrated absorbance, which evaluates both the area and the intensity of the positive staining. The density of circMET (the integrated optical density of all positive staining/total area) was calculated as representative of a particular sample, and the mean density was determined and used as a cutoff in subsequent analyses. The quantification of dipeptidyl peptidase 4(DPP4) was also performed via Image-Pro Plus version 6.2 software. For the CD8 T infiltrating lymphocytes in the TMA, positive cells in 200× images were counted in each section. For snail staining, the area of positive staining in an image was measured, and the average proportion (area of positive staining/total area) for each spot was calculated to represent a particular sample. The cutoff value was the mean of the data. The cutoff value for snail was 28% and the cutoff number of CD8+ tumor infiltrating lymphocytes was 30 in the 200× images.
Chromatin immunoprecipitation sequencing (CHIP-seq)
DNA from HCC cells was sonicated into 200–800 bp fragments and precipitated by centrifugation. DNA was resuspended in PBS and incubated with anti-Snail antibody (2–5 μg for 25 μg DNA) at 37 °C for 4 h. Samples were incubated with prewashed magnetic beads, collected with a magnetic frame, and washed with PBS. DNA was extracted with phenol chloroform, precipitated, and concentrated to prepare ChIP samples. DNA detection was performed by RT-PCR. Samples were constructed according to the Illumina library protocol and sequenced using the Illumina sequencing system.
Dual luciferase reporter assay
A mutant luciferase reporter vector was generated using a mutagenesis kit (Qiagen, CA, USA) according to the manufacturer’s instructions. Plasmids were transiently transfected into 293 T cells, which were lysed and collected after 48 h. Samples were centrifuged at 10,000–15,000 rpm for 3–5 min, and supernatants were collected.
Luciferase detection was performed according to the instrument instructions with a measurement time of 10 s and an interval of 2 s. For the assay, 20 μL of sample and 20 μL of firefly luciferase assay reagent were gently mixed 2–3 times, and relative light units (RLU) were measured with cell lysis buffer as a blank control. This process was repeated with Renilla luciferase assay reagent, and the degree of reporter gene activation was determined by using the ratio of both RLU values.
Chemokine chip
A chemokine chip assay was performed via a Raybiotech mouse cytokine array and the steps are listed in Additional file
4: supplementary materials and methods.
The in vivo tumor growth assays were performed using C57BL/6 mice. Mice were acquired from the Shanghai Institute of Material Medicine and were fed in a pathogen-free environment. Hep1–6 cells were injected subcutaneously into the mice with a 27-gauge needle. Tumor sizes were calculated by the following formula: volume (mm3) = [width2 (mm2) × length (mm)]/2.
Statistical analysis
Statistical analysis was performed with SPSS software (19.0, SPSS, Inc., Chicago, IL) as previously described [
4]. Values are shown as the mean ± standard deviation. The chi-square test and Fisher’s exact probability test were used for comparisons between groups. Correlation analysis was performed between circMET and Met. The recurrence and overall survival (OS) in HCC patients were analyzed using Kaplan-Meier’s method and the log-rank test. Cox’s proportional hazards regression model was employed to present the independent prognostic factors.
p < 0.05 was considered statistically significant.
Discussion
A large number of circRNAs have been successfully identified in diverse cells and tissues owing to advances in high-throughput deep sequencing [
35,
36]. Moreover, many circRNAs are expressed in a cell type-specific or tissue-specific manner, indicating that they might perform important biological functions [
16]. Dysregulation of circRNAs was recently shown to be involved in several pathological processes, including neurological disorders, cardiac hypertrophy and cancer [
36,
37]. By analyzing the expression of circRNAs in the 7q21-7q31 region, which was recently discovered to be a novel locus associated with both susceptibility to and prognosis of HBV-related HCC [
12], we found that circMET (hsa_circ_0082002) expression was upregulated in HCC tissues compared to paratumor tissues, and that high levels of circMET were related to poor prognosis of HCC patients.
Moreover, we found that high levels of circMET induce cell EMT and are associated with a tumor suppressive microenvironment in HCC. Chip-seq and luciferase reporter assays showed that circMET overexpression induced an immunosuppressive tumor microenvironment via the miR-30-5p/Snail/ DPP4/CXCL10 axis. Importantly, we demonstrated that the DPP4 inhibitor sitagliptin significantly improved the antitumor effect of PD1 in immunocompetent mice bearing tumors with high levels of circMET and Snail. Clinically, we determined that sitagliptin may enhance CD8+ T cell infiltration in HCC patients. Thus, we revealed that circMET is a new onco-circRNA that induces HCC development and immune tolerance via the Snail/ DPP4/CXCL10 axis. DPP4 inhibitors thus may be able to enhance the efficacy of checkpoint-inhibitor therapy in a subgroup of HCC patients.
Amplification of chromosome 7q21-7q31 is implicated in cancer susceptibility, recurrence and multidrug resistance, including in HCC [
12,
13]. Over the past decades, several oncogenes in this region, including
MET, HGF,
SHFM1 and
PEG10, have been revealed. Generally, a chromosomal aberration frequently harbors many cancer driver genes. According to previous reports [
28], we assessed the expression of several circRNAs in this region in HCC and corresponding paratumor tissues. circMET was significantly upregulated in HCC tissues and HCC cells with high metastatic potential. Moreover, circMET overexpression induced HCC cell EMT and shaped the tumor suppressive microenvironment. Thus, we provide evidence that circMET is another oncogene in the chromosome 7q21-7q31 region.
Our data provide direct evidence reinforcing the notion that EMT is associated with tumor immune inhibition, whereby Snail could serves as a transcription factor of DPP4, which induces local immunosuppression by negatively regulating lymphocyte trafficking via cleavage of the chemokine CXCL10 [
33]. As a Th1-type chemokine, CXCL10 plays an important role in determining effector T cell trafficking to the tumor microenvironment. Moreover, elevated CXCL10 in tumor cells can elicit potent tumor immunity, block cancer progression and enhance the clinical efficacy of immunotherapy [
38]. Our results showed that tumors with circMET overexpression had low levels of CXCL10 and less CD8
+ T cell infiltration than the control tumors. Moreover, we also showed a positive relationship between snail and DPP4, and a negative relationship between snail and CD8
+ T cells in human HCC tissues. The above results indicated that circMET and Snail were associated with the tumor immunosuppressive microenvironment. Importantly, we further determined that the administration of sitagliptin, a DPP4 inhibitor, could act synergistically with anti-PD1 treatment in immunocompetent mice; simultaneously, CXCL10 was upregulated by the sitagliptin administration. Thus, we revealed that the circMET/snail/ DPP4/CXCL10 axis is a vital mechanism in HCC immunosuppression, and identified a subgroup of patients who can benefit from sitagliptin administration. This suggests a role for DPP inhibitors in improving immunotherapy in a subgroup of HCC patients, which was consistent with previous publications in melanoma [
39]. Indeed, combined with the results of a retrospective cohort study conducted in HCC patients, we found that tissues from HCC patients with diabetes undergoing sitagliptin treatment had a higher level of CD8
+ T lymphocyte infiltration than tissues from HCC patients with diabetes without sitagliptin treatment, which further supported that sitagliptin could elevate the efficacy of anti-PD1 antibody therapy.
Generally, changes in the expression of circRNAs are associated with linear transcripts derived from the same gene, but many genes exhibit differentially regulated circRNAs and linear RNA variants. Here, we found that MET expression had no relationship with circMET expression. Interestingly, RNA-seq results indicated that HGF, SHFM1, and PEG10 were slightly downregulated in HepG2-circMET cells compared with HepG2-control cells, which we speculate may indicate a feedback inhibitory mechanism of circMET overexpression. Additionally, we also found that MET expression did not differ between cells with different circMET expression. Thus, the role of circMET in HCC progression is independent of MET function, which further supports the notion that circMET is a driver of HCC.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.