Introduction
Hepatocellular carcinoma (HCC) is responsible for the third most common cancer-related death worldwide [
1]. Importantly, the vast majority of HCC is caused by liver fibrosis or cirrhosis [
2]. Liver fibrosis occurs when inactive liver fibroblasts or hepatic stellate cells (HSCs) become activated after liver injury and turn into collagen-producing cells [
3]. In HCC, cancer-associated fibroblasts (CAFs) are loosely defined as HSCs found within the tumor mass [
4]. These CAFs have been implicated as etiologic players both in the cancer genesis and homeostasis. These data suggest that liver fibrosis, or activated HSCs, plays a crucial role in the development of HCC, which may be similar to the role of CAFs in desmoplastic tumors.
How fibroblasts are activated in HCC remains controversial. Recent studies have revealed multiple potential origins, including activation of stellate cells or portal fibroblasts or transdifferentiation of hepatocytes through epithelial-mesenchymal transition (EMT) [
5]. Stemmed from different origins, CAFs are highly heterogeneous, and they can be identified by different specific markers [
6]. Among them, α-smooth muscle actin (α-SMA) is the most commonly used marker for CAFs [
7]. Moreover, CAFs are believed to regulate the inflammatory microenvironment by expressing pro-inflammatory genes, such as IL-1β, IL-6, IL-8, TGF-β and CXCL12 as well as collagen [
8,
9]. The crosstalk between tumor cells and CAFs has been extensively studied [
10‐
12]. However, the mechanisms underlying the activation of HSCs by tumor cells remain largely unexplored in liver cancer.
An important type of cell-cell communication occurs through exosomes. These small, nanometer-sized (50–100 nM) vesicles of endocytic origin are released into the extracellular milieu by cells under physiological and pathological conditions, including antigen presentation and infectious agent transmission. Tumor exosomes are important mediators of the cross-talk between tumor cells and their microenvironment by sharing genetic information or functional proteins to modulate cellular behavior [
13,
14]. Tumor cell-derived exosomes are involved in the regulation of EMT, tumor angiogenesis, tumor metastasis and radioresistance [
15,
16].
It has been revealed that microRNA (miRNA) dysregulation greatly contributes to the activation of HSCs [
17]. Interestingly, there is a significant overlap in the list of up-regulated or down-regulated miRNAs between HCC tumor tissues and normal tissues [
18]. It has been demonstrated that secreted miRNAs can function in a paracrine manner in the surrounding microenvironment, promoting tumor development [
19]. Meanwhile, studies have indicated that exosomes contain a high level of miRNAs, and exosomal miRNAs have been shown to contribute to immunomodulation, chemoresistance and metastasis in multiple types of tumor [
20,
21]. However, it remains unclear how HSCs are activated through miRNA pathways.
In the present study, we aimed to classify the mechanisms underlying the activation of HSCs in liver cancer. Moreover, CAFs promoted a set of properties of vascular endothelial cells via increasing the secretion of angiogenic factors. The bilateral interaction between primary tumor cells and stromal cells further illuminated a new mechanism of tumor progression and offered new opportunities for potential therapeutic strategies targeting HCC development.
Methods
Specimens and primary cells
Human serum specimens and liver tissues were collected from healthy donors and HCC patients before or after resection in the Nanjing Drum Tower Hospital in Nanjing, China. All procedures were approved by the Ethical Committee of the Nanjing Drum Tower Hospital. Written informed content was obtained from every participant prior to study.
Cell culture
The human liver cancer cell lines (97H, LM3 and Huh7), the vascular endothelial cell line HUVEC, and the HSC cell line LX2 were purchased from Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Shanghai Institute of Cell Biology) and maintained in DMEM (Gibco) supplemented with 10% FBS (Gibco). The normal liver cell line LO2 was purchased from Cell Bank of Type Culture Collection of the Chinese Academy of Sciences and maintained in RPMI-1640 medium (Gibco) supplemented with 10% FBS (Gibco). All cell lines were cultured at 37 °C in a humidified incubator containing 5% CO2. Cell lines were authenticated by short tandem repeats (STR) profiling and confirmed to be mycoplasma negative.
Reagents and antibodies
Antibodies against CD63 (ab125011, 1:1000), α-SMA (ab32575, 1:100), FAP (ab207178, 1:1000), FSP (ab124805, 1:1000), GAPDH (ab8245, 1:10,000), CD9 (ab92726, 1:1000), CD81 (ab79559, 1:1000), FASN (ab22759, 1:500), ATP citrate lyase (EP704Y, 1:1000) and phospho-ATP citrate lyase (T447/S451) (ab53007, 1:1000) and USP2 (ab66556, 1:500) were purchased from Abcam (Cambridge, MA, USA). Antibodies against phosphor-PTEN (Ser380/Thr382/383) (9554S, 1:1000), total PTEN (7960 T, 1:1000), PDK1 (3062 T, 1:1000), phospho-PDK1 (Ser241) (3438 T, 1:1000), phospho-Akt (Ser473) (4060 T, 1:1000) and AKT (4691 T, 1:1000) were supplied by Cell Signaling Technology (Beverly, MA, USA). AKT inhibitor MK-2206 was provided by Selleck (Houston,USA).
Western blotting analysis
Whole-cell protein extracts were homogenized in lysis buffer and centrifuged at 12,000 r.p.m. for 15 min. Protein concentrations were determined by bicinchoninic acid (BCA) assay. After immunoblotting, the proteins were transferred onto nitrocellulose membranes, followed by incubation with specific antibodies. The immunocomplexes were then incubated with the fluorescein-conjugated secondary antibody, and immunoreactive bands were visualized by an Odyssey fluorescence scanner (Li-Cor, Lincoln, NE).
RNA interference
siRNAs and mimics of indicated miRNAs were obtained by RiboBio Company (Guanngzhou, China). The sequences of siRNAs and miRNA mimics referred above were listed in Additional file
1: Table S1. Transfection with siRNAs and miRNAs was completed using riboFECT™ CP (RiboBio) according to the manufacturer’s instructions.
Animal studies
To examine the roles of exosomes in CAFs, 1 × 106 Huh7 cells were intravenously injected into male nude mice through the tail vein (Chinese Science Academy, Shanghai, China). Subsequently, mice were randomly divided into groups and intravenously injected with equal numbers of exosomes from different tumor cells twice a week for 1 month. For xenograft assays, 1 × 106 Huh7 cells and 1 × 106 LX2 cells were injected subcutaneously into the right side of each male nude mouse (Chinese Science Academy). The sizes of tumors (length × width: 2 × 0.5) were measured at the indicated time points, and tumors were obtained at 4 weeks after injection. All animal experiments were approved by the University Committee on Use and Care of Animals of Nanjing Drum Tower Hospital.
Cell viability assay, migration assay and wound-healing assay
For cell viability assay, Cell Counting Kit 8 (CCK-8) assay (Dojindo Laboratories, Kumamoto, Japan) was used to assess cell viability according to manufacturer’s instructions. For migration assay, 5 × 104 LX2 cells were plated into 24-well transwell plates with inserts (Corning). The medium in inserts was free of FBS, whereas the medium outside the inserts was supplemented with 10% FBS. To detect exosome function, equal quantities of tumor-derived exosomes were added into the inserts. After 24 h, the cell inserts were fixed and stained according to manufacturer’s protocols. Representative fields were photographed, and the number of migrated cells per field was counted. For wound-healing assay, equal numbers of LX2 cells were plated into six-well plates. Then the cell monolayers were wounded with a pipette tip to draw a gap on the plates. After treated with tumor-derived exosomes, fibroblasts that migrated into the cleared section were observed under microscope at the specific time points.
RNA extraction and qRT-PCR
HSC RNA was extracted from snap-frozen liver tissues with TRIzol™ reagent (Life Technologies, USA) according to the manufacturer’s instructions. Briefly, cells in six-well plate were homogenized in 1 mL TRIzol™ reagent at room temperature. Then 200 μL trichloromethane was added and mixed thoroughly, followed by centrifugation at 12,000×g for 15 min at 4 °C. The upper clear phase was collected and added with 500 μL isopropanol, followed by centrifugation at 12,000×g for 10 min at 4 °C. The liquid supernatant was discarded, and the pellet was washed with 75% ethanol. The RNA was re-suspended in 50 μL RNase-free DEPC-water and stored at − 80 °C.
Reverse transcription was performed with PrimeScript™ RT Master Mix (Takara, Japan) according to the manufacturer’s instructions. qRT-PCR was performed using TB Green™ Premix Ex Taq™ (Takara, Japan) on an ABI PRISM 7500 real-time PCR System (Applied Biosystems, USA). Primers used for qPCR are shown in Additional file
1: Table S1. The relative expression levels of mRNAs were calculated with 2
–ΔΔCt method. GAPDH was selected as the housekeeping gene.
miRNeasy Mini Kit (Qiagen) was used to extract RNA from exosomes in plasma/medium according to the manufacturer’s instructions, and cel-miR-39 (Takara) was added into each sample at a final concentration of 10 pmol/μL acting as external reference. Total RNA was stored at − 80 °C for subsequent experiments.
Reverse transcription and qRT-PCR for exosomal miRNA, as well as internal reference U6 were performed using miRNA RT-PCR Quantitation Kit (Qiagen) according the manufacturer’s instructions. Briefly, after an initial denaturation step at 95 °C for 3 min, the amplifications were carried out with 40 cycles at a melting temperature of 95 °C for 15 s, and an annealing temperature of 62 °C for 34 s. The relative expression levels of exosomal miRNAs were calculated with 2–ΔΔCt method.
Oil red staining
Cells were fixed in 10% formalin, washed with 60% propylene glycerol, and then stained with 0.5% oil red O (Sangon Bio) in propylene glycerol for 10 min at 60 °C. The red lipid droplets were visualized by microscopy.
Immunohistochemistry and in situ hybridization analysis
For immunohistochemistry, the slides were incubated with above-mentioned primary antibodies, followed by incubation with horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology). Finally, the staining processes were performed with diaminobenzidine colorimetric reagent solution (Dako, Carpinteria, USA) and hematoxylin (Sigma Chemical Co., USA). For in situ hybridization analysis, hsa-miRNA-21 miRCURY LNA detection probe (Exiqon, Denmark) was used, and the total staining processes were carried out according to manufacturer’s protocols. Images were captured with Aperio ScanScope AT Turbo (Aperio, USA) and assessed with image-scop software (Media Cybernetics, Inc.).
Immunofluorescence analysis
Immunofluorescence analysis was performed according to previously established protocols. Cells were seeded into 24-well dishes and fixed by 4% paraformaldehyde 24 h later. Fixed cells were stained with α-SMA and FAP antibodies (mentioned above), followed by incubation with FITC-conjugated anti-mouse IgG and Cy3-conjugated anti-rabbit IgG (Abcam). Representative images were detected by fluorescence microscopy (Leica, German), and data were processed via Photoshop.
EC matrix was thawed at 4 °C overnight, and required wells of pre-chilled 96-well plates were coated with 50 μL diluted EC matrix and incubated at 37 °C for 1 h to solidify. Subsequently, 150 μL of HUVEC cells (1 × 104) with different conditional media were added to the solidified matrix and incubated at 37 °C for 12 h. Endothelial cell formation was observed using a microscope. Focus was placed on distinct areas, and the tubes formed were counted according to the kit procedure.
Isolation and analysis of exosomes
For exosome isolation, equal numbers of different cells were transplanted into 10-cm plates and maintained in fresh DMEM supplemented with serum, which was depleted of exosomes by centrifugation at 12,000×g overnight. After 48 h, CM was collected and filtrated through 0.22-μm filters (Millipore, USA). Exosomes in CM or serum samples were isolated by ultracentrifugation according to the standard methods described previously [
22]. Ultracentrifugation experiments were performed with Optima MAX-XP (Beckman Coulter, USA). Exosomes were observed by Philips CM120 BioTwin transmission electron microscope (FEI Company, USA).
Exosomes tracing
For exosome-tracing experiments, tumor cells were pre-treated by PKD67 (Beyotime, China), and exosomes in CM were obtained as described above. After incubation with recipient cells that were pre-treated with FAP antibody and DAPI (Beyotime), exosomes were observed by fluorescence microscope (Leica, German).
Transmission electron microscopy
The obtained exosomes were fixed with fixative buffer containing 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M PBS. After embedding, samples were cut into 0.12-μm sections and stained with 0.2% lead citrate and 1% uranyl acetate. The images were detected by a JEOL TEM-2000 EX II (JEOL, Tokyo, Japan).
Cell cycle analysis
HSCs were seeded into six-well plates at a density of 1.5 × 105 cells/well overnight, and the cells were conditionally cultured for 24 h. Cell cycle was detected using cell cycle analysis kit (Thermo Fisher Scientific, USA) according to the manufacture’s protocols and measured by flow cytometry.
Collagen contraction assay
A total of 2 × 105 HSCs were suspended in 100 μL DMEM. Then the cell suspension was mixed with 100 μL of collagen mix containing 68.75 μL DMEM, 0.72 μL 1 N NaOH and 31.25 μL type 1 rat tail collagen (Corning), added to one well of 6-well plates and allowed to solidify for 45 min at 37 °C. After incubation with medium containing HCC-derived exosomes, the gels were photographed by digital camera.
Statistical analysis
Data analysis was performed using the SPSS software version 16. Each experiment was carried out in triplicate at least, and all results were presented as mean ± SD. χ2-Test and Student’s t-test were used to assess statistical significance. Kaplan–Meier analysis and log-rank tests were applied for survival analysis. A p value of < 0.05 was considered as statistically significant.
Discussion
As a dynamic system orchestrated by intercellular communications, tumor microenvironment is responsible for tumor progression and metastasis. Therefore, it is necessary to study the interaction between tumor and stroma mediated by exosomes. In our study, we first analyzed the different profiles of exosomal miRNAs between HCC cells and normal liver cells. Then, we identified that miRNA-21 was directly transferred from tumor cells to HSCs in tumor parenchyma via exosomes, and miRNA-21 could convert HSCs to CAFs by down-regulating its target PTEN to activate PDK1/AKT signaling pathway. In addition, CAFs promoted tumor development by angiogenesis through secreting IL-6 and IL-8. The crosstalk between tumor cells and HSCs further elucidated the molecular mechanism of HCC invasion, and explained why liver cancer was highly invasive. Furthermore, our data indicated that high miRNA-21 expression in serum exosomes was correlated with low survival rate, holding important implications for efficient prevention and therapeutic strategies. The involvement of miRNAs in cancer shows that the expressions of several miRNAs are dysregulated in neoplastic tissues [
33]. The identification of several targets of miRNAs, which are actually classical oncogenes or tumor suppressors, has led to the widely accepted idea that miRNAs play pivotal roles in cancer initiation, progression and metastasization [
34]. In another elegant study employing K-ras G12D non-small-cell lung cancer (NSCLC) mouse model, the results show that incidence of lung tumors is significantly high in miRNA-21-overexpressing mice. Consistent with this finding, deletion of miRNA-21 has resulted in suppression of Kras-driven transformation in vitro and tumor development in vivo [
35]. Our data demonstrated that tumor-derived exosomal miRNA-21 converted HSCs to CAFs in HCC. Moreover, stimulation of exosomal miRNA-21 from HCC cells had a positive correlation with tumor volume in nude mice. Meanwhile, high miRNA-21 expression in HCC predicted a poor outcome. Previous studies have mainly focused on miR-21 overexpression in cancer cells, which promotes cellular proliferation, evasion of apoptosis, EMT and invasion. However, more and more attention has been paid to the role of miR-21 in CAFs. In a study on colorectal cancer, miRNA-21 expression is increased in stromal cells compared with normal tissues, and the ectopic expression of miR-21 drives the trans-differentiation of fibroblasts into myofibroblasts and increases invasion in vitro [
36]. MiRNA-21 expression in lung fibroblasts may trigger fibroblast trans-differentiation into CAFs, supporting cancer progression. Furthermore, patients with miR-21-high CAFs exhibit lower survival compared with those with miR-21-low CAFs [
37]. In summary, CAFs may further secrete exomal-miR21 to promote HCC progression. Future work will be required to study the role of miRNA-21 in liver cancer.
PTEN stands for phosphatase and TENs in homolog deleted on chromosome 10, and it is a classical tumor suppressor gene located in the 10q23 region of chromosome 10 encoding for a 403-amino acid multifunctional protein (predicted MW 47 kDa), which possesses lipid and protein phosphatase activities [
38,
39]. PTEN functions as a classical tumor suppressor, and it is mainly involved in the homeostatic maintenance of the AKT cascade [
40]. PI3K, a lipid kinase activated by receptor tyrosine kinases, G protein-coupled receptors and RAS activation, converts the lipid second messenger phosphatidylinositol 4,5-bisphosphate (PIP2) to phosphatidylinositol 3,4,5-trisphosphate (PIP3). PIP3 recruits PDK1 and AKT to the plasma membrane, where AKT is phosphorylated on Thr308 by PDK1. By dephosphorylating PIP3 to PIP2, PTEN reverses the action of PI3K, thereby hampering all downstream functions controlled by the AKT pathway, such as cycle progression, lipid synthesis and stimulation of angiogenesis. We found that decreased PTEN expression disturbed by HCC cell-derived exosomes increased phosphorylation levels of PDK1 and AKT, regulated lipid metabolism and promoted the release of angiogenic substances (VEGF, MMP2, MMP9, bFGF and TGF-β). PTEN also controls cell-cycle progression by decreasing the level of cyclin D1 in the nucleus and regulates cellular senescence. Consistent with previous studies, our results indicated that HSCs with low expression of PTEN under the stimulation of HCC cell-derived exosomes also showed high level of cyclin D1 and increased proportion of cells in the S phase.
Of the factors that were found to be induced by CAFs after co-culture with HCC cell-derived exosomes, the up-regulation of VEGF, MMP2, MMP9, bFGF and TGF-β was especially relevant, since these factors are known to promote cancer growth, invasion and angiogenesis through autocrine or paracrine signaling [
41‐
43]. Additionally, these cytokines are associated with advanced stages of breast cancer and a poor clinical outcome. VEGF can also stimulate angiogenesis and it is significantly associated with a poor survival [
44]. MMP2 and MMP9 are known multifunctional proteins. Data from diverse experimental models have indicated that these proteases affect cellular activities, including proliferation and survival, gene expression as well as multiple aspects of inflammation. Previous study has reported associations between high expression of MMP2 or MMP9 and tumor aggressiveness in liver cancer [
45]. As a highly angiogenic molecule, bFGF is of particular interest since it is capable of promoting both the proliferation and migration of endothelial cells in various tumor models, and it can function synergistically with other factors to promote angiogenesis [
46]. In summary, we found that HCC cell-activated CAFs could induce the expressions of several factors related to angiogenesis and tumor progression in HCC.
Aberrant activation of lipogenesis is a dominant oncogenic event in human HCC. Importantly, no significant differences were detected in the extent of de novo lipogenesis with regard to HCC etiology, suggesting that exacerbated lipogenesis was a general molecular phenotype in hepatocarcinogenesis. Indeed, previous reports have demonstrated that both hepatitis B and C viruses are able to induce FASN expression [
47,
48], and over-expression of FASN is a typical feature of liver cancer under another predisposing condition, the alcoholic steatohepatitis [
49]. Furthermore, a rat model of insulin-induced hepatocarcinogenesis is characterized by strong up-regulation of FASN [
50], which resembles the occurrence of HCC in human affected by type II diabetes mellitus and/or metabolic syndrome, two clinical conditions associated with an increased risk of liver cancer development [
51]. Our data suggested that abnormal lipid metabolism was present not only in cancer cells but also in stromal cells. CAFs also expressed a high level of FASN with its maintenance protein USP2a and phosphorylated ACLY. Abnormal lipid accumulation was also coupled with the activation of HSCs.