Background
Hepatocellular carcinoma (HCC) is the third leading cause of cancer mortality worldwide [
1]. Primary HCC lesions can be removed completely when detected at an early stage, but intrahepatic recurrence of HCC and extrahepatic metastasis are very frequent, giving rise to a poor prognosis for patients [
2,
3]. It is widely accepted that both differentiated hepatocytes and cells with progenitor characteristics, known as cancer stem cells (CSCs), can cause HCC [
4‐
7]. Forty percent of HCCs are clonal, and potentially derived from progenitor/stem cells. Moreover, these cells have a critical role in the development and progression of HCC [
8]. Liver CSCs have been isolated from primary HCC specimens and patients’ sera as circulating cells, and from HCC cell lines by use of surface antigens [
9‐
11]. CD90, the epithelial cell adhesion molecule (EpCAM) and CD133 have been found to recognize three distinct cell populations that differ from one another in features and behavior in determining cancer phenotypes [
12].
CD90 (Thy-1) is a 25-37 kDa glycophosphatidylinositol (GPI)-anchored protein expressed by several cells such as T-cells, neurons, endothelial cells and fibroblasts. It is involved in cell-to-cell and cell-matrix interaction, apoptosis, adhesion, migration, fibrosis, and cancer development [
13]. Concerning the liver, the expression of CD90 has been linked to hepatic stem/progenitor cells [
14] and, during tumor growth, it has been correlated with an aggressive phenotype [
15], and associated with low differentiated HCC and poor prognosis [
16‐
18]. CD90+ CSCs obtained from HCC cell lines, from tumor tissues and peripheral blood as circulating cancer cells displayed, in contrast to the other CSC populations, a mesenchymal phenotype and, most importantly, a greater capacity to metastasize when injected into immunodeficient mice [
11,
12,
19]. Moreover, recent data associate CD90 expression with early HCC recurrence [
20]. Gene expression and miRNA analysis in CD90+ HepG2 cells have revealed an imbalance in the expression of apoptotic and anti-apoptotic genes compared with CD90 negative cells [
21]. However, we are still far from understanding the molecular mechanisms underlying the more aggressive and metastatic phenotype of these cells compared with the other liver cancer cells.
Tumor development is dependent on the reciprocal interactions between cancer cells and the surrounding microenvironment. It is well known that in addition to pathways involving cell-to-cell contact and the release of soluble factors, cancer cells are able to communicate with the tumor microenvironment (e.g., myeloid cells, fibroblasts, endothelial cells) through the intercellular exchange of proteins and genetic materials via exosomes [
22].
Exosomes are spherical membrane vesicles of endocytic origin, with an average size of 40 to 150 nm [
23], released by both normal and diseased cells after the fusion of multivesicular bodies with the plasma membrane. First considered as collectors of cellular waste materials, exosomes have assumed a leading role in the regulation of the tumor microenvironment. Depending on their content, exosomes can affect tumor cells and surrounding stroma by influencing major cellular pathways, such as apoptosis, cell differentiation, angiogenesis, and metastasis [
24]. These vesicles act as cargos that release bioactive molecules e.g., lipids, proteins, and nucleic acids in target cells. Interestingly, recent observations have identified a vesicle–mediated transfer of lncRNAs as an important mechanism in the development of HCC [
25]. In this paper we demonstrate that CD90+ cells, derived from HCC cell lines, release exosomes that, in turn, are able to influence endothelial cells by promoting angiogenesis and stimulating their adhesive properties. Furthermore, our results suggest the lncRNA H19 as a possible mediator of angiogenic effects.
Discussion
CD90+ liver CSCs have been found in primary tumors, and circulating in the blood of HCC patients, and are associated with early recurrence, metastasis, and poor prognosis [
19,
20]. Our study highlights the ability of CSC-like CD90+ cells, but not hepatoma cells, to influence endothelial cell phenotype through the release of exosomes.
In a solid tumor, the CSC’s niche is composed of an extracellular matrix (ECM), mesenchymal stem cells, tumoral cells, immune cells, and endothelial cells, all of which converge in determining the fate of CSCs through extracellular signals [
38]. Little is known about the modulation of the tumor microenvironment by CSCs. Several studies have described exosomes as signaling extracellular organelles that modulate the tumor microenvironment, promoting angiogenesis and tumor progression [
27,
39]. Our data indicate that exosomes released by CSC-like CD90+ liver cells are able to promote an angiogenic phenotype in cultured endothelial cells. CD90 + -derived exosomes induced in HUVECs an increase in the production and secretion of VEGF, the most powerful pro-angiogenic cytokine, as well as of its receptor VEGF-R1. This increase was accompanied by an amplification in the number and length of tube-like structures formed by HUVECs in culture.
It is abundantly documented that metastatic processes induce changes in the endothelial surface antigens, with an increase in adhesion molecules, which, in turn, favor the adhesion and the consequent intra- or extra-vasation of metastatic cells. We found that exosomes released by CD90+Huh7 cells, and not by hepatoma cells, increased the number of HUVECs expressing ICAM-1 and, more extensively, increased the adhesion between endothelial cells and the CSC-like CD90+ cells. Our data also indicate that the CD90+ released exosomes may be able to promote metastasis.
Recently, Patel et al. demonstrated that lncRNA could be selectively packaged in extracellular vesicles released by hepatoma cell lines and transported to other cells, with subsequent modulation of cellular function [
25,
40]. LncRNA are emerging as molecular players in several biological processes acting at epigenetic, transcriptional and post-transcriptional levels or processing small non-coding RNAs [
41]. H19 was among the first lncRNAs to be identified and studied principally for its monallelic expression, and as regulator of IGF2 abundance [
42,
43]. As already described for other lncRNAs, H19 can work as a microRNA sponge, miRNAs precursor, or epigenetic modulator [
44,
45], and has been found overexpressed in several tumors, and able to promote tumor growth [
46,
47] and progression [
47,
35,
36]. Concerning the liver, H19 has been clearly involved in hepatocarcinogenesis [
48] and hepatic metastases [
49]. Several indications correlate H19 with angiogenesis [
50,
51]. Northern analysis has indicated a high expression of H19 during development of rat aorta that decreases in differentiated tissue and, interestingly, re-appears following vascular injury
in vivo and
in vitro [
50], though no observations of the overexpression of H19 in endothelial cells have been published.
In this study, we demonstrate, for the first time to our knowledge, that H19 is highly expressed in a subpopulation of hepatoma cells that expose the surface antigen CD90 and are characterized, by others, as CSC-like cells [
11,
12,
15,
29]. We found that CD90+Huh7 cells package lncRNA H19 inside exosomes, thus delivering it to possible target cells. Exosomes released by CD90+ liver cancer cells could be internalized by endothelial cells, influencing these in a pro-metastatic way. Moreover, we identified in H19 an important player of this process. H19 overexpression in endothelial cells is able to up-regulate the VEGF production and release, increase the ability of HUVEC cells to arrange
in vitro tubular-like structures, and promote heterotypic adhesion between endothelial cells and CSC-like liver cells. Silencing experiments revealed LncRNAH19 as the principal player of the exosome-mediated VEGF increase, while suggested the presence of other molecular actors that, transported or induced by CD90 + -derived exosomes, and together with H19, affect endothelial cells in a pro-metastatic way. However, the mechanisms of action through which this lncRNA controls an endothelial phenotype remain to be elucidated.
Material and methods
Cell culture and reagents
Human umbilical vein endothelial cells (HUVECs) were obtained from Lonza (Verviers, Belgium) and grown in endothelial growth medium (EGM, bullet kit, Lonza) according to supplier’s instructions. Huh7 cells and Sk-Hep cells were cultured in DMEM medium (Euroclone, UK), and supplemented with 10 % fetal bovine serum (Euroclone, UK), 2 mM L-glutamine, 100 U/ml penicillin and 100 mg/ml streptomycin (Euroclone, UK).
Sorting CD90+Huh7 cells
Huh-7 human hepatocellular carcinoma cells were stained with anti-CD90 PE (BD Pharmingen™ 555596), and surface marker was determined by flow cytometry. CD90+ and CD90- cells were sorted through a FACSAria I (BD Biosciences). A purity check was done after the sorting by re-running a small fraction of the sorted populations. All cells showed over 85 % purity.
Immunocytochemistry
Immunocytochemistry was done on PFA 4 % fixed cells, and stained with the following antibodies: the primary antibodies were anti-E-Cadherin (BD Biosciences 610181), anti-HNF4a (Abcam ab41898), and anti-Vimentin (Epitomics, 2707-1); the secondary antibodies were Alexa-Fluor 488 and Alexa-Fluor 594, from Molecular Probes. The nuclei were stained with NucRed® Live 647 (Catalog number: R37106, Life Technologies), and preparations were analyzed by confocal microscopy (Leica TSC SP8).
Exosome preparation and characterization
Huh7, CD90+ Huh7 and Sk-Hep cells were grown with 10 % ultracentrifugated FBS, and conditioned medium was collected 48 h after culture; exosomes were subsequently isolated by serial centrifugation [
26]. Briefly, culture medium was centrifuged subsequently for 5 min at 300 × g, 15 min at 3,000 × g, 30 min at 10,000 × g and ultracentrifuged 90 min at 100,000 × g in a Type 70 Ti, fixed angle rotor. Peletted exosomes were washed and then resuspended in PBS. Exosome protein content was determined with the Bradford assay (Pierce, Rockford, IL, USA). On average we recovered 10 micrograms of vesicles from 25 ml of conditioned medium from 3 × 10
6 cells. The intensity autocorrelation functions of diluted vesicle samples were measured by dynamic light scattering (DLS) using a Brookhaven Instruments BI-9000 correlator and a BI200-SM goniometer, equipped with a solid-state laser tuned at 532 nm. The size distribution was determined from the vesicle diffusion coefficients by standard analysis [
52]. Thirty μg of protein for each sample, exosomes, and cells, were analyzed by western blot for Alix (3A9-Cell Signaling Technology #2171S),) Tsg101 (Santa Cruz Biotechnology sc-7964) and HSC70 (Santa Cruz Biotechnology sc-7298).
Uptake of exosomes by HUVECs
Exosomes from Huh7, CD90+ Huh7 and SkHep cells were labeled with PKH26 according to supplier’s instructions, suspended in low serum medium (5 μg/ml), and incubated with HUVECs for 1, 3, and 6 h at 4° or 37 °C. After incubation, cells were processed as previously described [
26].
HUVECs treatment
HUVECs were grown at a density of 100.000cells/well in a 12 wells plate, and treated for 18 h with 5 μg/ml of exosomes in low serum medium; untreated cells were considered control. Plasmid for psiCHECK2-H19 and the Empty vector psiCHECK2 (kindly provided by Dr Y. Huang [
45]]), H19 siRNA (SR319206B Origene Technologies) and scramble negative control (SR30004 Origene Technologies) were transfected in HUVECs with Attractene Transfection Reagent (cat.number.1051531, Quiagen) following manufacturer’s indications.
RNA extraction and real-time PCR
RNA was extracted using the commercially available illustra RNAspin Mini Isolation Kit (GE Healthcare), according to manufacturer’s instructions. Total RNA was reverse-transcribed to cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystem). RT-QPCR was done in 48-well plates using the Step-One Real-Time PCR system (Applied Biosystem). Real-time PCR was performed in duplicates for each data point. For sybr-green method the oligonucleotide used were β-actin for5’-ATCAAGATCATTGCTCCTCCTGA-3’rev 5’CTGCTTGCTGATCCACATCTG-3’; H19 for5’-GCACCTTGGACATCTGGAGT-3’rev5’-TTCTTTCCAGCCCTAGCTCA-3’, VEGF for5’-CGAGGGCCTGGAGTGTGT-3’rev5’-CGCATAATCTGCATGGTGATG-3’, VEGF-R1 for5’-CGGTCAACAAAGTCGGGAGA-3’rev5’-CAGTGCACCACAAAGACACG-3’, VE-CADHERIN for5’-GATCAAGTCAAGCGTGAGTCG-3’ rev5’-AGCCTCTCAATGGCGAACAC-3’. VCAM1, ICAM, H19 and β-actin transcript levels were measured by TaqMan Real-Time PCR using the TaqMan gene expression assay: Hs00174239_m1, HS 00277001_m1, Hs00262142_g1 and Hs99999903_m1, respectively (Life Technologies,). Changes in the target mRNA content relative to housekeeping were determined with the ΔΔct Method.
HUVECs were seeded at 50,000 cells/well in growth factor-reduced Matrigel-coated 24 well plate and incubated up to 2 h at 37 °C. Tube formation was examined under an inverted microscope and photographed at 20× magnification. The length of the cables was measured manually with IMAGE-J software (
http://rsbweb.nih.gov/ij/).
FACS analysis
Two hundred thousand (200,000) cells were washed in PBS and incubated with 0.5 μg ICAM-1-FITC (sc-107, Santa Cruz). Viable cells were gated by forward and side scatter, and analyzed on 100,000 acquired events for each sample. Samples were analyzed on a Partec CyFlow Space using the Partec FloMax® software.
Adhesion assay
In order to evaluate the ability of CD90+ Huh7 cells and SkHep cells to adhere to HUVECs, an adhesion assay was performed, as previously described [
26].
ELISA
HUVEC conditioned medium was collected 18 h after exosome treatment or transfection with pH19 or pEmpty. VEGF concentrations were quantified using the ELISA kit (KHG0111, LifeTechnologies), according to manufacturer’s protocol.
Array for long non-coding RNA
In order to study lncRNA expressed in the sorted population, a LncProfiler lncRNA qPCR array was performed (System Bioscience) on Huh7, CD90 + Huh7 cells and their exosomes following manufacturer’s indications. After amplification, ΔΔct of CD90 + Huh7 was normalized on ΔΔct of Huh7, and data were expressed as fold induction of the sorted population compared with the parental cells.
Statistical analysis
In vitro experiments were repeated three times, giving reproducible results. Data are presented as mean values ± standard deviation (SD) of three independent experiments. Statistical analysis was done using Prism 4 (GraphPad Software Inc., San Diego, CA, USA); one-way ANOVA (non-parametric) was performed, followed by Dunnett’s multiple comparison test.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AC, RA, VC, ALD contributed to the conception and design of the study. AC, RA, VC, ALD, LS, SB, SR, MM, RA contributed to the generation, collection, assembly, analysis and/or interpretation of data. AC, RA, VC, ALD, MT, CM, FD, GDL, contributed to drafting or revision of the manuscript. All authors approved the final version of the manuscript.