Background
During her lifetime, one in eight women will be diagnosed with breast cancer and despite the multimodal therapeutic schemes available, metastatic disease remains the main cause of mortality. Over half of those that succumb will have overt liver metastases [
1]. Metastasis is a multistep process that requires detachment of the cancer cells from the local tumor, extravasation and survival in the blood circulation, followed by an ectopic survival and proliferation at a distant site. The initial step for cells to leave the primary tumor is the loss of epithelial cell–cell adhesions and subsequent acquisition of a more mesenchymal-like invasive phenotype, referred to as Epithelial–Mesenchymal Transition (EMT) [
2], main hallmark of which is loss of E-cadherin, a calcium (Ca
2+)-dependent transmembrane glycoprotein that mediates cell-cell cohesion. However, we and others have shown that while necessary for invasion, E-cadherin downregulation is reversed during the metastatic seeding [
3‐
5] and that entry into dormancy relies on this cancer cell phenotypic plasticity through which highly dedifferentiated breast cancers appear more epithelial in the metastatic milieu [
2,
6‐
8]. Our laboratory has focused on the development and use of an all-human liver microphysiological system (MPS) that uses fresh human hepatic tissue to recapitulate the hepatic niche (HepN) so as to study the steps of the metastatic seeding, tumor outgrowth and entry into dormancy of cancer cells as a model that closely mimics the process in vivo and allows the monitoring and immediate interactions between the cancer cells and the hepatic microenvironment [
9‐
11].
How cancer cells communicate with their local and distant environment is undergoing a re-evaluation with the discovery that all cells release extracellular vesicles (EVs) to their surrounding milieu. EVs vary in size and function from apoptotic bodies (1000–5000 nm) to microvesicles (MVs 200–1000 nm) and exosomes (30-150 nm). Exosomes were first described in the early 1980s during reticulocyte maturation and were believed to be a form of “cellular waste disposal”. It took several decades for the scientific community to realize that the exosomes are critical to tissue homeostasis and master regulators of cell-to-cell communication [
12,
13]. Contrary to the other EVs formation process, the biogenesis of exosomes is intricate and requires multiple levels of regulation, including the formation of late endosomes or multivesicular bodies (MVBs). MVB formation is primarily regulated by the endosomal sorting complexes required for transport (ESCRT) and during this stage, cytoplasmic RNA and proteins get incorporated into exosomes. Subsequently these MVBs fuse with the cell membrane and release exosomes to the extracellular space [
14]. All exosomes contain common proteins that serve as exosome markers during their isolation, including the tetraspanins CD9, CD81 and CD63, ESCRT associated proteins, like Alix and TSG101 and cytoplasmic proteins, such as Hsp70. However, the cell of origin and its physiologic state at any given time will also be reflected in the exosomal components, in addition to those structural ones noted above.
Through their cargo DNA, mRNA, miRNA, enzymes and soluble factors, exosomes can mediate various autocrine and paracrine signals and orchestrate intercellular communication. More specifically tumor derived exosomes or oncosomes [
13] have largely been described as promoters of tumor progression and drivers of metastasis by generating a premetastatic milieu through fibroblast activation, immune suppression, extracellular matrix remodeling, enhanced vasculogenesis and/or drug resistance [
13,
15‐
17]. Even though our understanding of the oncosomal contribution during tumor progression has advanced during the last decade, all our information is extracted from 2D cultures that do not closely mimic the metastatic cascade as it unfolds in the human body. Moreover, even though normal exosomes have been shown to be important for tissue homeostasis and wound healing, very little is known about the effects of normal tissue exosomes on tumor cells, particularly in reference to distant seeding and tumor outgrowth during cancer progression.
Herein, we studied the effect of exosomes in the metastatic seeding and proliferation of breast cancer cells to the liver. We showed that priming of the HepN with breast cancer derived exosomes results in significant enhancement of the cancer cell seeding in the liver, followed however by a suppressed tumor cell outgrowth in proliferation experiments. The same data was recapitulated in our Liver MPS that closely mimics the metastatic process. We hypothesized that the suppressed proliferation was due to the effect of the HepN derived exosomes on the colonizing breast cancer cells. We validated that the HepN significantly contributes to the exosome pool and that these exosomes, significantly different in size, protein and miRNA content, through miRNA regulation, activate pathways that suppress cancer cell proliferation and invasion. More importantly we show that treatment of MDA-231 and MDA-468 with purified normal HepN derived exosomes increases E-cadherin and ZO-1 protein expression levels in both breast cancer lines. Our data identifies a novel mechanism of regulation of the metastatic cascade, introducing the normal tissue/HepN derived exosomes as significant modulators of MErT during seeding and suppression of tumor growth once the breast cancer cells have reached the liver.
Methods
Cell lines and culture
Breast cancer cell lines, MDA-231 and MDA-468, were cultured in RPMI medium. MDA-231 obtained from ATCC were stably transfected with Red Fluorescent Protein (RFP) expressing vector as previously described [
8]. Prostate cancer line DU145 was cultured in DMEM medium. The melanoma line WM35 was maintained in MCDB153: L15 4:1 medium mixture with addition of 5% Fetal Bovine Serum (FBS), 5 μg/ml insulin and 2 mM CaCl
2. The WM852 melanoma line was cultured in DMEM:L15 3:1 mixture with addition of 10% FBS [
18]. TPF16238, melanoma cells derived from patient sample, were kindly provided by Dr. Kirkwood and maintained in RPMI with 1 mM L-glutamine, 0,1 mM NEEA and 10 nM HEPES. All cells were incubated at 37 °C in 5% CO
2, and supplemented with 1%
v/v penicillin/streptomycin, and 10% v/v FBS, unless specified otherwise (Gibco, Life Technologies, Grand Island, NY, USA). The Trypan Blue exclusion method was used to assess cell viability. As FBS contains exosomes of bovine origin, all cell cultures were depleted of serum for 24 h prior to exosomes isolation as previously described by
Théry et al. [
19].
Liver cells
The primary human hepatocytes (Hep) and non-parenchymal cells (NPCs) were obtained from therapeutic partial hepatectomies for metastatic colorectal carcinoma or, more usually, benign diseases such as focal nodular hyperplasia and hemangiomas. The cells are available from the NIDDK-funded Liver Tissue and Cell Distribution System (LTCDS) with the procurement core directed by Dr. David Geller at the University of Pittsburgh and funded by the NIH (Contract #HHSN276201200017C). The livers are perfused and separate isolations of Hep and NPCs were provided to us, as previously described [
20]. We further process the NPC fraction (to eliminate contaminating debris, hepatocytes, and red blood cells) as previously reported [
21].
Exosome isolation
Exosomes were purified from cell culture supernatants by ultracentrifugation as previously described [
15]. Briefly, FBS free culture medium was collected and centrifuged at 300×g for 10 min to remove whole cells. The supernatant was then centrifuged at 3,000×g for 20 min to remove dead cells and debris. This supernatant was centrifuged at 10,000×g for 30 min to further remove cell debris. This supernatant was then spun at 100,000×g for 70 min and the pellet was washed with excess PBS to remove contaminating proteins followed by a 70 min centrifugation at 100.000×g to obtain the exosome pellet.
Isolation of exosomes from the liver MPS was performed using the Total Exosome Isolation Reagent from cell culture media (Life Technologies); this method allowed for more efficient handling of smaller volumes from the MPS. After a 20 min centrifugation at 3,000×g the supernatant, containing the exosomes, was removed and combined with 1× volume of the Total Exosome Isolation Reagent and incubated overnight at 4 °C. The exosomes were harvested after a 60 min centrifugation step at 10,000×g. The exosome pellet was subsequently washed in Phosphate Buffered Saline (PBS) followed by a 70 min spin at 100.000×g. A bicinchoninic acid (BCA) protein assay kit (Pierce, Thermo Fisher, OH, USA) was used to determine the concentration of exosome proteins and performed as per the manufacturer’s instructions.
Transmission electron microscopy
5 μl of freshly isolated exosomes in PBS suspension were applied to copper mesh Formvar coated carbon stabilized grids. They were allowed to adsorb to the grid for 2-3 min and then were wicked off with filter paper. For negative staining of the exosomes, 1% Aqueous Uranyl Acetate (5-10 μl) was applied to the grid for 30 s, then wicked off with Whatman filter paper. Grids were allowed to thoroughly dry before viewing.
Exosome staining-functional assay
Freshly isolated exosomes were stained with the red lipophilic dye DiI (Thermo Fisher, OH, USA) which is incorporated in the outer exosome membrane. Exosomes were stained in the dark at room temperature in 1 μM DiI, then washed in PBS and centrifuged for 70 min at 100,000×g to remove the unincorporated dye. Pelleted exosomes were resuspended in PBS and added to the cells in culture.
Exosome RNA isolation, cDNA and miRNA analysis
After the last PBS wash the exosome pellet was resuspended in RLT buffer and total RNA isolation was performed using the RNeasy kit (Qiagen, Hilden, Germany). RNA concentration was measured using the NanoDrop2000 Spectrophotometer (yield range: 60-160 ng/μl). For miRNA analysis, the Cancer Focus microRNA PCR Panel, miRCURY LNA was used (Exiqon, Denmark) by following the manufacturer’s protocol, and using the ExiLENT SYBR® Green master mix. Prior to amplification RNA was converted to cDNA with the Universal cDNA Synthesis Kit II (Exiqon, Denmark).
Plasmids and reagents
The pCT-CD63-GFP vector was obtained from System Bioscience (SBI, Palo Alto, CA, USA) and MDA-231 RFP+ cells were transfected using Lipofectamine2000 (Life Technologies, Grand Island, NY, USA) using 2,5 μg of plasmid DNA.
Ex vivo liver MPS
The ex vivo Liver MPS is assembled as recommended by the manufacturer (CN Bio Innovations Ltd., Oxford, UK). Scaffolds for tissue growth are freshly coated with 1% rat tail collagen type I (BD Biosciences, Life Technologies, Grand Island, NY, USA) in PBS for 1 h at 37 °C and washed with PBS before placement in the system. Prior to seeding the system is primed with 1% Bovine Serum Albumin (BSA) at 37 °C, which is then replaced with William’s Medium E (Gibco, Life Technologies) supplemented with the Hepatocyte Thawing and Plating Supplement Pack (Life Technologies). To recapitulate physiologic conditions hepatocytes and NPCs are plated at a 1:1 ratio (6 × 105 cells per well) in the system. Hepatocytes are cultured for approximately 16 h in the plating William’s E medium, described above, and are subsequently changed to William’s E medium supplemented with the Hepatocyte Maintenance Supplement Pack (Life Technologies), prepared following the manufacturer’s instructions. The medium is exchanged completely every 48 h. On day 2 and day 3 the system is primed using cancer derived exosomes (20 μg/well/day) and on day 3 cancer cells are introduced into the formed liver tissue. For seeding into the Liver MPS, 500 MDA-231 RFP+ cells were trypsinized and neutralized in complete growth medium, centrifuged and resuspended in hepatocyte maintenance medium.
2D co-culture experiments
6 × 105 primary human hepatocytes were plated using William’s Medium E supplemented with the Hepatocyte Thawing and Plating Supplement Pack in a well of a 6-well plate coated with 1% rat tail collagen in PBS and were allowed to attach overnight. The following day the medium was changed to William’s Medium E supplemented with the Hepatocyte Maintenance Supplement Pack. On day 2 and day 3 the cultures received cancer derived exosomes (20 μg/well/day) for priming and on day 3, 500 RFP+ cancer cells were seeded onto the hepatocyte monolayers. For the transwell assay, after priming of the HepN on day 2 and day 3, 2 × 104 cells were plated on the transwell coculture inserts (Millipore).
Immunofluorescence microscopy/ cancer cell detection and enumeration
MDA-231 and MDA-468 breast cancer cells were seeded on coverslips precoated with 1% collagen and then fixed with 2% paraformaldehyde in PBS. Following cell permeabilization with 0.1% Triton X-100, non-specific epitopes were blocked with 2% BSA and the cells were incubated with primary antibody diluted in 0.5% BSA at 4 °C overnight. After 2 washes in 0.5% BSA, cells were stained with Alexa Fluor® 488 conjugated secondary antibody at room temperature for 1 h. DAPI was applied to stain the nucleus.
For the 2D co-culture experiments, cells were fixed on day 4 and day 7 with 2% paraformaldehyde in PBS for 1 h at 4 °C and then were visualized using an Olympus BX51 with a 4× (PlanApo NA.0.08) objective (Olympus America Inc.). Digital images were obtained on an Olympus CCD camera using the SPOT 5.2 image acquisition software. The level of RFP-expressing breast cancer cells was determined using the ImageJ software. Images from the entire scaffolds were inclusively thresholded and the RFP+ portion measured as a percentage of the total area.
Western blot
Exosomes and cells were lysed using radioimmune precipitation assay (RIPA) buffer with phosphatase inhibitor mixture II, III (Sigma) and protease inhibitor mixture (BD Biosciences). Protein extracts, separated by SDS-PAGE and transferred onto PVDF membranes, were probed with antibodies against E-cadherin (BD Biosciences), ZO-1 (BD Biosciences), CD81 and CD63 (System Bioscience) or actin (Sigma). Proteins of interest were detected with HRP-conjugated secondary anti-mouse IgG antibody (Sigma, 1:10.000) and visualized with the Pierce ECL Western blotting substrate (Thermo Scientific, Rockford, IL), according to the provided protocol. The Exo-Check™ exosome antibody arrays were purchased from System Bioscience and performed as instructed by the manufacturer.
Cell cycle analysis
Samples were collected and fixed in 70% ethanol overnight at 4 °C. For cell cycle analysis, fixed cells were washed in PBS and treated with RNase for 20 min at 37 °C. 5 μg/mL of Propidium Iodide (PI) were added to each sample and analyzed by flow cytometry.
Statistical analysis
Statistical analysis for all experiments was performed with using Prism7 software (GraphPad, La Jolla, CA). An independent Student’s t-test was used to determine statistical differences between experimental and control values. P values <0.05 were considered statistically significant.
Discussion
Metastasis is a multistep complex process that despite having detrimental effects on the patient’s prognosis is inherently highly inefficient. In order to metastasize, the cancer cells need to undergo EMT to escape the primary tumor, extravasate into the circulation, avoid cell death due to lack of support from the surrounding cancer mass and extracellular matrix and subsequently colonize the distant site and resume proliferation. Many of these steps have been well studied and elucidated but recent advances have added a new level of complexity by identifying the cancer derived exosomes as critical shuttles for cell to cell communication, promoting tumor progression and increasing the efficacy of the metastatic cascade [
13]. Tumor derived exosomes mediate the transfer of oncogenic proteins and nucleic acids thus inducing a persistent and efficient modulation in the genotype and phenotype of the recipient cells [
34‐
36]. Furthermore, they promote the recruitment and reprogramming of the tumor microenvironment to form a pro-tumorigenic soil, by establishing a positive feedback loop that shapes the ever-evolving tumor microenvironment and allows cancer cell seeding, proliferation and dissemination at distant sites [
37‐
39]. However, the specific mechanisms through which healthy cells are triggered to release exosomes that promote or suppress the malignant behavior of cancer cells remain to be determined.
It must be noted that while we find classical exosomes, and our preparations appear mainly to consists of these as noted by classical shape and size on electron microscopy, the isolation techniques will enrich for all extracellular vesicles, both the larger blebs and the smaller exomeres. Herein we are not distinguishing between these, but rather using exosomes as the shorthand to encompass all such communicating vesicles.
Using our Liver MPS to closely reproduce the events of the metastatic cascade (ref BJC and LoC papers), we show that priming of the hepatic microenvironment with cancer derived exosomes promotes the seeding of breast cancer cells to the metastatic site. This is in agreement with published literature that shows the effect of the oncosomes in establishing a pre-metastatic niche [
37‐
39]. However, more importantly we were able to recognize the significant contribution of the normal HepN derived exosomes to the exosomal pool during metastasis. It should be noted that the source of liver cells comes from donors undergoing partial hepatectomies for both benign and malignant diseases (though the cells are obtained from tissue distant from the pathology). While we do not note differences in MPS behavior between these sources, it is possible the pathology necessitating the partial hepatectomy influences the exosome profile and cargo even weeks later in the MPS. The HepN derived exosomes differ from the cancer derived exosomes in their size and their miRNA and protein cargo. These HepN derived exosomes are produced constantly but become more relevant once the cancer cells reach the liver microenvironment as they are now within range of interaction. It is also important to keep in mind that the cancer cells could potentially affect the normal cells through exosomes and soluble factors, modifying the exosome pool that the latter secrete. Herein we show that after the initial seeding to the liver, the cancer cells communicate with the HepN through exosomes and via miRNA transfer, the hepatic microenvironment suppresses cancer cell proliferation and promotes slow cell cycling.
These studies are performed ex vivo which entails both advantages and limitations. The all-human situation cannot be recapitulated in animal models, and thus select signals or miRNA that do not cross species may be approachable only with such a MPS. Further, the isolation of compartments and tissues, with rapid sampling of unmodified effluent is not readily achieved in vivo. Still, once specific molecules are identified as causally required, these may be targetable in animal models of spontaneous metastasis in future investigations.
Dormancy, defined as a more quiescent cell state, marked by slow proliferation and low metabolic activity has proven to be the main cause of treatment failures and cancer relapses. IPA of the miRNA profile of the MDA-231 in HepN shows a significant suppression in the pathways involved in cancer cell proliferation and invasion indicating that the interaction with the liver microenvironment results in alterations in the MDA-231 expression profile and changes in the cancer cell phenotype. To corroborate these findings we performed cell cycle analysis of 2D co-cultures and showed a delay of the cells in the G0-G1 phase and that the HepN derived exosomes are enriched in miRNAs involved in the regulation of G1 arrest. In the exosome protein array we observe a significant elevation of the exosomal marker CD81 in the MDA-231 cells in the HepN. The tetraspanin CD81 has been reported to be important regulator of cell fate in hematopoietic stem cells (HSC) and CD81 can drive proliferating HSC to quiescence [
40]. The HepN derived exosome-induced suppression of cancer cell proliferation is potentially a “tumor suppressive” mechanism that the normal tissue microenvironment at the distant site utilizes in order to maintain a steady state and minimize the changes that occur within it after the arrival and seeding of the cancer cells, however in the long term this could have detrimental effects for the patient since the cells can resume proliferation and outgrow at any time in the future. This observation has significant clinical implications considering that the cell cycle status often determines the cell’s response to chemotherapeutic agents. Future studies are required to further characterize the genotypic and phenotypic changes that occur and to further validate the dormant state of the cancer cells and their potentially acquired chemoresistance.
Furthermore we have previously shown that interaction of the breast cancer cells with the hepatocytes and NPCs in the liver leads to a re-expression of E-cadherin and a partial MErT [
3,
8]. Our initial studies had found that neither conditioned media alone nor hepatocyte-derived matrix could trigger E-cadherin re-expression in this breast carcinoma line, though the combination of the two was noted to lead to a weak re-expression of E-cadherin [
8]. Herein we provide evidence that this effect on the breast cancer cells is exosome mediated and that treatment of breast and prostate cancer cells with purified normal HepN derived exosomes increases E-cadherin and ZO-1 protein expression levels and induces a more epithelial-like morphology. No significant changes were observed in the levels of the mesenchymal marker Vimentin suggesting a partial MErT. Our findings highlight the importance of the exosomes derived from the receptive microenvironment in the phenotypic plasticity of the cancer cells during tumor progression and the metastatic cascade. Through MErT and expression of adhesion molecules, the reversible phenotypic plasticity allows cancer cells to adapt to the foreign soil during ectopic organ colonization [
41]. Expression of the epithelial-marker and the cell adhesion molecule E-cadherin on breast cancer cells may be another mechanism to facilitate adhesion to hepatocytes [
24].
The effects that we observe could largely be attributed to the exosomal transfer of miRNAs as these are master regulators. By comparing the miRNA content of MDA-231 that were exposed to HepN derived exosomes to MDA-231 we see that the miRNAs that are significantly upregulated or downregulated have been reported previously to regulate E-cadherin expression and MErT in various cancer types [
28‐
36]. Additional testing of our miRNA hits in the top and bottom bins is essential for the identification of E-cadherin and MErT transcriptional regulators in our system since targeting with antagomirs could be a useful clinically applicable method to target the downstream pathways. Furthermore it is known that in most carcinomas, and namely in the MDA-231 breast cancer line, E-cadherin appears to be turned off at the translational level by promoter hypermethylation [
42,
43]. Aberrations of DNA methylation have been inferred to play an important role during the initiation and progression of various cancers [
44] and they could potentially play a critical role in our system as well since exosomes can act as epigenetic reprogrammers [
45]. Definitive assignment of critical role of likely multiple and overlapping miRNA will require both defining the cell types of origin and then targeted disruption, both of which lie beyond the scope of the present missive.
Conclusions
Our data collectively identifies a novel mechanism of regulation of the metastatic cascade, introducing the normal tissue/HepN derived exosomes as significant modulators of MErT during seeding and suppression of tumor growth once the breast cancer cells have reached the liver. The importance of the normal exosomes during the metastatic cascade has been widely underestimated, however we show that they become critical players during metastasis. We highlight the presence of a well-orchestrated exosome mediated bidirectional crosstalk between the cancer cells and the HepN that differentially regulates the steps of the metastatic cascade. The importance of the exosomes as signal shuttles stems from their newly established close proximity with the cancer cells. The concentration of exosomes in the blood are likely too dilute to strongly signal the liver from the primary site, and particularly in reverse. However, once the disseminated cell intravasates into the metastatic locale, the microenvironmental concentrations of these signals would be magnified, and likely more effective in altering cellular and tissue phenotypes. The present study posited such a directed bidirectional interaction. For the first time we introduce evidence that not only shows that cancer derived exosomes modify the premetastatic niche but the metastatic niche in turn also modulates the subsequent steps of the metastatic cascade determining the cancer cell fate.