Role of mesenchymal stromal cell-derived extracellular vesicles in tumour microenvironment.

Stromal cells, deriving from mesenchymal stromal cells (MSCs), are crucial component of tumour microenvironment and represent key regulators of tumour processes. MSCs can be recruited to the tumour environment and interact with many cellular elements, thus influencing tumour biology. Cell-to-cell communication is in part mediated by the release of extracellular vesicle (EVs). EVs can induce significant molecular changes in recipient cells, delivering bioactive molecules. In this review, we describe the MSC-derived EVs content and discuss their role in different processes related to cancer biology. Furthermore, we summarize chemical or biological EVs modifications aiming to develop more efficient antitumor therapies.


Introduction
Tumour initiation and progression require the generation of a supportive niche that constitute a favourable microenvironment promoting tumour cell viability, proliferation, and invasion. Different cell types contribute to the tumour cell niche including stromal cells, endothelial cells and immune cells. Stromal cells represent the main cell component with both supportive and immunoregulatory functions; they derived from multipotent cells of mesodermal origin, called mesenchymal stromal cells (MSCs), which virtually reside in all tissues [1]. Therefore, stromal cells can be recruited to the tumour environment and establish dynamic interactions with tumour cells and other cellular elements [2][3][4]. The role of MSCs in tumour microenvironment has been increasingly studied in the last years. MSCs have shown to enhance or suppress tumour progression and metastasis both in vitro and in vivo depending not only on tumour model and stage considered, but also on the dose and time of cell treatment [5,6].
Recently, a new mechanism of intercellular communication within the tumour microenvironment has been recognized. This mechanism involves the release of membrane-derived vesicles, called extracellular vesicles (EVs). EVs include exosomes, the smallest EV fraction arising from intracellular endosomes, and microvesicles (MVs) that are generated by budding from the plasma membrane. EVs influence various biological processes by both activating directly into target cell surface receptors through protein and bioactive lipid ligands, and delivering different effectors, including transcription factors, oncogenes, mRNAs, and non-coding regulatory RNAs (lncRNAs), thus inducing functional changes in recipient cells [7]. The pattern of proteins, lipids, and nucleic acids contained in EVs depends on their cell origin and may be influenced by different environmental conditions. EVs are released by either tumour cells or their surrounding cells and are considered as important mediators of the cell cross-talk in the tumour microenvironment. Tumour-derived EVs are involved in the immune response regulation, thus creating an immunosuppressive niche and triggering immune escape mechanisms [8]. Furthermore, tumour-derived EVs contribute to the formation of metastasis by inducing persistent molecular changes in stromal components, promoting angiogenesis and epithelial-mesenchymal transition (EMT) [8]. On the other hand, EVs derived from MSCs (MSC-EVs) can regulate cell apoptosis, proliferation, invasion of tumour cells and influence several tumour processes by delivering their content in recipient cells (Fig. 1). Although the role of tumour-EVs is rather well-defined, MSC-EVs, as the originating cells, can act on tumour progression with opposite effect and their role has not been clearly elucidated yet. In this review, we report recent findings on the role of MSC-EVs in tumour biology, focusing on their involvement in tumour growth, angiogenesis and drug resistance (Table 1). addition to their structural role, these lipids have a broad effect on signal transduction and regulation known tumour suppressor, potentiating signalling pathways that drive apoptosis and cell cycle arrest [11,12]. On the other hand, alterations in ceramide turnover favour the opposite effect promoting cell growth [13]. Furthermore, cancer cells are associated to low level of sphingomyelin. The specific activation of sphingomyelin synthases has been shown to lead to cell cycle arrest, cell differentiation and apoptosis in cancer cells [14]. Diacylglycerol kinases, a family of enzymes catalysing the transformation of diacylglycerol into phosphatidic acid, are often up-regulated in many tumour cell types, suggesting that the reduction of diacylglycerol could promote tumour immune-escape [15]. Taken together, these data open an important scenario on the role of lipid transfer mediated by MSC-EVs on tumour processes. Moreover, MSC-EVs contain various bioactive factors that could be involved in several pro or anti-tumour processes, including transcription factors, oncogenes, and nucleic acids, such as mRNAs and small and lncRNAs. MSC-EVs content could be strictly affected by the tissue of origin. Through high-throughput techniques, Vallabhaneni et al. described several known tumour supportive factors in bone marrow-MSC-EVs (BM-MSC-EVs), such as PDGFR-beta, TIMP-1, and TIMP-2 [11]. BM-MSC-EVs showed also the expression of miRNA-21 and miRNA-34a that are involved in cell survival and proliferation [11,16]. MVs released by human umbilical cord-derived MSCs (UC-MSCs) have shown to express many angiogenesis-promoting molecules, including IL-6, bFGF, and UPAR. MSC-MVs also contain angiogenin, VEGF, MCP-1 and the receptor-2 for vascular endothelial growth factor [17]. Furthermore, under hypoxic condition, the expression of some of these cytokines are higher than under normoxic condition [17]. Recently, several works explored the protein expression profile of exosome or MV-enriched fraction derived from MSCs, thus identifying a wide spectrum of proteins associated to inflammation, Wnt signalling pathway components and several putative paracrine effectors of angiogenesis [18,19]. Among the molecules contained in MSC-EVs, many functionally active mRNAs and non-coding RNAs involved in the crosstalk between tumour cells and MSCs have been described [20]. Identified miRNAs can be classified in families or clusters with different functions. Let-7 and miR-15-16 families released by MSC-EVs mainly possess anti-tumour activity, targeting oncogenes such as RAS, MYC, HMAGA2, and LIN28, and the angiogenic factor VEGF-A, respectively [21]. MiR-221 and miR-222 are also involved in angiogenic mechanisms and they have been identified in EVs released by rat MSCs [22]. The expression of lncRNAs has been increasingly studied in human cancer cells and more recently in EVs. In human BM-MSC-EVs the expression of lncRNA involved in cell proliferation, such as lnc-7SK, has been found, thus suggesting their possible role in tumour regulation [11].    Recently, Dong et al [23]. showed in vivo a pro-tumour activity of EVs derived from UC-MSCs on lung adenocarcinoma (LUAD) growth; in fact, proliferation and apoptosis of LUAD cells were supported by the EV-mediated transfer to cancer cells of miR-410 that directly targets PTEN. MVs released by human Wharton's Jelly MSCs (WJ-MSCs) displayed the same supportive activity on renal cancer cell growth and diffusion both in vitro and in vivo, promoting the progression of cell cycle from G0/1 to S in tumour cells [24]. In this system, the MSC-MVmediated transfer of RNA induced HGF synthesis in tumour cells, thus activating AKT and ERK1/2 signalling. The involvement of ERK1/2 signalling was also previously reported by Zhu et al., who showed the activation of such pathway in a human cancer cell line (SGC-7901) treated with MSC-exosomes [25]. Moreover, tumour incidence in mice implanted with SGC-7901 cells or a human colon cancer cell line (SW480) mixed with BM-MSCs or MSC-exosomes was similar and significantly higher than in mice implanted with tumour cells alone [25]. On the other hand, the inhibitory effect of MSC-EVs on tumour growth has been suggested by various reports. MSC-EVs reduced cell cycle progression in HepG2 hepatoma, Kaposi's sarcoma, and Skov-3 ovarian tumour cell lines, while they induced apoptosis in HepG2 and Kaposi's cells and necrosis in Skov-3 [26]. Furthermore, BM-MSC-exosomes promoted dormancy in metastatic breast cancer cells [27]. Such effect has been correlated to the down-regulation of MARCKS expression in cancer cells mediated by the EV-transfer of miR-23b. Human WJ-MSC-EVs induced cell cycle arrest and apoptosis on bladder tumour T24 cells in vitro and in vivo, through the reduction of phosphorylated AKT protein kinase and the increase of cleaved caspase-3 [28]. Moreover, exosomes secreted by human Adipose Tissue-derived MSCs (AT-MSCs) reduced cell viability, wound-repair capacity, and colony formation ability of A2780 and Skov3 ovarian cancer cells [29]. In particular, A2780 cells responded dose-and time-dependently following AT-MSCexosomes treatment and they showed a strong upregulation of several pro-apoptotic signalling molecules, including phosphorylated p53, BAX, activated Caspase9, and activated Caspase3, in association with a down regulation of the anti-apoptotic BCL2 protein [29]. The controversial data on the effect of MSC-EVs on tumour growth highlight the complexity of their functions, that could be even opposite depending on several factors, such as the type and the tissue origin of MSCs; in addition, cancer cells may show different responses depending on the tumour model considered. The results can be also influenced by the method of EV isolation: some authors purified exosomes or MV-enriched fraction, while others did not purify the two classes of EVs. Moreover, the method of MSC culture may affect the experimental results. Serum-deprivation, a cell culture system often used for EVs isolation, can modify the expression of important molecules involved in cell survival and apoptosis, such as miR-21 and miR34a, which in turn can affect the release and content of EVs [11]. In the context of multiple myeloma, opposite effects of MSC-exosomes derived from either normal individuals or patients were observed on tumour cell growth. MSC-EVs derived from patients induced tumour growth, whereas MSC-EVs derived from healthy donor showed an anti-tumour effect [30]. Furthermore, the role of MSC-EVs could be different according to the pathological stage of the tumour. Vallabhaneni et al. showed that MSC-EVs supported primary breast tumour progression, but reduced metastasis formation through the UBE2N/Ubc13 pathway [31]. Therefore, tumour model and stage as well as MSC culture conditions are all factors that may influence the release of EVs, their content, and therefore their effects on tumour growth.

MSC-EVs and angiogenesis
The influence of EVs released by various cell types on vascular development, growth, and maturation has been widely documented and reviewed by Todorova et al. [32]. The role of EVs on angiogenesis has been assessed on both endothelial cells, evaluating their effect in modulating vascular structures, and tumour cells, studying the expression of pro-and anti-angiogenic factors [33].
Angiogenesis plays a crucial role in numerous physiological and pathological responses. Tumour development requires blood vessels for oxygen and nutrients supply, which are both essential factors for tumour growth. Tumour cells frequently overexpress pro-angiogenic factors, such as VEGF, strictly associated with cancer progression [34]. MSCs contribute to this process by controlling blood vessel density and functions [35]. As already underlined for tumour cell survival and growth, the effect of MSCs on angiogenesis have yielded contradictory results, including both anti-angiogenic and pro-angiogenic effects [36,37]. Although the evidence of strong pro-angiogenic properties correlated with the secretion of soluble factors and EVs by tumour components, the effect of MSC-EVs on tumour angiogenesis remains relatively unclear [32]. Murine MSC-EVs may shuttle anti-angiogenic molecules into tumour cells, such as miR-16, a miRNA known to target VEGF, leading to the inhibition of angiogenesis both in vitro and in vivo models of breast cancer [38]. On the contrary, Zhu et al. clearly demonstrated that BM-MSC-exosomes strongly increase in vitro the expression of CXCR4 and VEGF in human gastric carcinoma (SGC-7901) and colon cancer (SW480) cell lines [25]. These authors observed that MAPK pathway, including ERK1/2 and p38, and -in minimal part -JNK pathway, were responsible for the increase of VEGF and CXCR4 after MSC-exosomes treatment. Furthermore, MSC-exosomes strongly induce the expression of α-SMA, CXCR4, VEGF and MDM2 in in vivo [25]. In breast cancer cells, BM-MSC-derived exosomes may induce a significant reduction of VEGF expression and secretion by modulating the mTOR/ HIF-1α signalling axis and affecting the vascular behaviour of endothelial cells in vitro [39]. In addition, VEGF down-regulation has been associated with the horizontal transfer of miR-100 mediated by MSCexosomes, which has been further confirmed by the ability of anti-miR-100 to rescue the inhibitory effects of MSC-derived exosomes [39]. Liang et al. showed the opposite effect mediated by exosomes secreted by human Adipose Tissue-derived MSCs (AT-MSCs) [33]. In this experimental model, MSC-exosomes promoted angiogenic processes in vitro by inducing the upregulation of the pro-angiogenic genes Ang1 (ANGPT1) and Flk1 (KDR) and the downregulation of the anti-angiogenic factors Vash1 and TSP1 in endothelial cells (HUVECs) [33]. MSCexosomes may also increase the number of vascular structures in vivo. These observations correlated with the transfer of miR-125, mediated by MSC-exosomes, targeting the angiogenic inhibitor delta-like 4 (DLL4) [33]. Many other reports show that MSC-EVs contain various miRNAs probably involved in modulating angiogenesis, including miR-30b, miR-222, miR-21 and let-7f, and miR148a [40,41]. Furthermore, McBride et al. showed the reduced ability of BM-MSC conditioned medium to promote angiogenesis after CD63+ exosome depletion [42]. The pro-angiogenic activity was also shown in exosomes isolated from UC-MSCs, with a mechanism dependent on Wnt4/β-catenin [43]. In response to hypoxia, EVs isolated from AT-MSCs stimulated angiogenesis in vivo [44]. The ability of MSC-EVs to modulate angiogenic processes was assessed in other contexts different from tumours. Considering their pro-angiogenic properties, MSC-EVs have been proposed as novel therapeutic approaches for cardiovascular diseases as well as brain, renal, and retinal injuries [45].

MSC-EVs and drug resistance
Drug resistance of cancer cells is a multifactorial process due to dynamic interactions between tumour cells and the surrounding microenvironment. This process includes many mechanisms, such as defective drug accumulation, alteration of drug targets and cellular pathways and, eventually, lack of induction of tumour cell death [46]. Tumour drug resistance mediated by microenvironmental stimuli has been clearly underlined by several reports and reviewed in the context of different tumours [47]. Tumour-EVs can mediated drug resistance through several mechanisms, including drug sequestration, delivery of specific RNA molecules and proteins, or through the well-known crosstalk between cancer cells and MSCs [48]. MSC-exosomes containing miR-222 and miR-223 may promote quiescence and drug resistance in breast cancer cells [49]. Furthermore, Ono et al. demonstrated that BM-MSC-exosomes decreased the sensitivity of breast cancer cells to docetaxel [27]. UC-MSC-exosomes significantly induced drug resistance of gastric cancer cells both in vivo and ex vivo [50]. MSC-exosomes inhibited 5-fluorouracil-induced apoptosis and promoted the expression of various multi-drug resistance-associated proteins, such as MDR, MRP and LRP. These findings have been associated to the activation of the CaM-Ks/Raf/MEK/ERK pathway mediated by MSC-exosomes [50]. Furthermore, MSC-EVs may induce drug resistance also in haematological tumours. BM-MSC-exosomes conferred resistance to bortezomib in multiple myeloma cells by influencing several pathways, including p38, p53, c-Jun N-terminal kinase, and Akt pathways [51]. BM-MSC-EVs also increased the chemoresistance to several drugs commonly used for chronic lymphocytic leukemia (CLL) B-cells, including fludarabine, ibrutinib, idelalisib and venetoclax [52]. The treatment with MSC-EVs significantly reduced the number of apoptotic CLL B-cells only after 24 h. Notably, EVs completely suppressed apoptosis induced by cladribine and bortezomib [52]. In addition, rat MSC-EVs may protect rat pheochromocytoma cells from the cytotoxicity induced by glutamate, which induce the reduction of Bcl-2 expression with the concomitant increase of Bax and cleaved caspase-3 expression. This effect was significantly reversed by the pretreatment with MSC-MVs, suggesting their protective effect in pheochromocytoma cells [53].

MSC-EVs and drug delivery
Pharmaceutical research has recently focused some strategies to design drug carriers that could be used for delivering drugs specifically to the site of interest [54]. Liposomes are considered an effective drug delivery system based on nanotechnology science and may have numerous therapeutic applications [55,56]. To date, at least 19 therapeutic approaches with liposomal drug delivery systems have been approved by FDA and EMA (www.fda.gov; www.ema.europa.eu/ema/ ), and many phases I-III open interventional studies are in progress. However, these formulations show several drawbacks related to stability, fatty acyl moiety oxidation, drug loading, liposome targeting, and drug leakage and release [57]. Considering the similarity between liposomes and EVs, the strategy to develop EV-based drugs has been recently proposed. Compared to liposomes, EVs possess some advantages in terms of clearance and immunogenicity. In addition, EV membranes are structurally comparable to cell membranes, thus favouring their biocompatibility and a lower toxicity [57]. Because of their ability to deliver bioactive factors among cells, one of the most intriguing therapeutic potential of EVs is the possibility to load them with the desired drug. EVs can be theoretically loaded with both hydrophilic and lipophilic drugs and, interestingly, they can deliver them to specific tissues, including the brain, considering their capacity to cross the blood-brain barrier, where they have a high stability [58]. Experimentally, there are two different approaches for EV drug loading; i.e. exogenously or endogenously. Endogenous loading can be carried out by strong overexpression of the drug molecule in releasing cells, while exogenous loading requires the collection of extracellular vesicles before their loading with the desired drug by using different manipulation, such as electroporation [45]. Pascucci et al. successfully obtained MSC-EVs loaded with Paclitaxel, showing their in vitro efficacy against pancreatic cancer cell growth [59]. Additionally, Del Fattore et al. demonstrated that MSC-EV loading with Vincristine promoted cytotoxicity toward glioblastoma cells, with a higher effect as compared to both the free drug and the unloaded EVs [60]. The possibility to directly engineer MSCs to secrete EVs enriched of specific components with anti-tumour activity is emerging. Lou et al. reported encouraging results in the context of hepatocellular carcinoma. Here, AT-MSCs were transfected with a miR-122 expression plasmid, leading to the release of miR-122 enriched exosomes. Such exosomes have shown to increase the chemosensitovity of hepatocellular carcinoma cells in vitro by enhancing cell apoptosis and cell cycle arrest, especially when HepG2 cell line was considered [61]. Moreover, intra-tumour injection of miR-122 enriched exosomes sensitized hepatocellular carcinoma cells to sorafenib in vivo [61]. Considering that breast cancer cell lines stably expressing miR-379 showed a reduced capability in terms of tumour formation and growth, O'Brien et al. proposed MSC-EVs enriched with miR-397 as an innovative therapy for metastatic breast cancer [62]. They have engineered human BM-MSCs through lentiviral transduction to release EVs enriched with miR-379 for in vivo therapeutic approach. The treatment with cell-free EVs loaded with miR-379 had a therapeutic effect with no adverse effects, while the administration of MSC expressing miR-397 did not interfere with tumour growth [57]. The same experimental approach was previously reported by Yuan et al. using the pro-apoptotic TNF-related apoptosis-inducing ligand (TRAIL) [63]. TRAIL and its receptors were used as targets of several anti-cancer drugs, but the clinical application of recombinant TRAIL showed several difficulties in terms of bioavailability and resistance of cancer cells. These problems could be fixed by using MSC-EVs as TRAIL vehicle. Yuan et al. developed TRAIL-enriched MSC-EVs and assessed their cancer cell-killing efficacy [63]. TRAIL enriched MSC-EVs induced apoptosis in 11 cancer cell lines in a dose-dependent fashion. They also demonstrated that caspase activity inhibition or TRAIL neutralisation abolished the cytotoxicity mediated by TRAIL-enriched EVs. Furthermore, the combined treatment of TRAIL-enriched MSC-EVs and CDK9 inhibitor enhanced the induction of apoptosis in TRAIL-resistant cancer cells [63]. Despite these encouraging results, further pre-clinical studies are necessary to clinically harness MSC-EVs as drug vehicle in the context of tumours. Moreover, their therapeutic application is still hampered by the lack of standardized and reproducible isolation and drug loading methods. In addition, understanding which population of EVs is most suitable for delivering drug molecules is also required for future therapeutic approaches.

Conclusions
The bidirectional transfer of molecules between tumour cells and tumour microenvironment plays a crucial role in cancer development. In this context, MSC-EVs represent potential vehicles for the delivery of pro-or anti-tumorigenic molecules. By transferring their content into recipient cells, MSC-EVs significantly influence cancer cell survival, growth, angiogenesis and drug resistance, even though some results are still controversial because of the lack of conformity in the assessment of biological functions. We discussed such controversial data specifying relevant experimental information, including the effect mediated by MVs or exosomes and MSC tissue sources. Nevetheless, standardized protocols for isolation, quantification and characterization of EVs are still required to clarify their role in tumour microenvironment and to aim for potential clinical applications. Furthermore, because of the capacity of EVs to encapsulate molecules, including proteins and nucleic acids, therapeutic strategies based on engineered EVs are currently on study and the preliminary results are quite promising, thus supporting encouraging prospects for cancer treatment.