Extracellular vesicle-mediated cell–cell communication in haematological neoplasms

Abstract

Crosstalk between bone marrow tumour cells and surrounding cells, including bone marrow mesenchymal stromal cells (BM-MSCs), endothelial cells and immune cells, is important for tumour growth in haematological neoplasms. In addition to conventional signalling pathways, extracellular vesicles (EVs), which are endosome-derived vesicles containing proteins, mRNAs, lipids and miRNAs, can facilitate modulation of the bone marrow microenvironment without directly contacting non-tumourous cells. In this review, we discuss the current understanding of EV-mediated cell–cell communication in haematological neoplasms, particularly leukaemia and multiple myeloma. We highlight the actions of tumour and BM-MSC EVs in multiple myeloma. The origin of EVs, their tropism and mechanism of EV transfer are emerging issues that need to be addressed in EV-mediated cell–cell communication in haematological neoplasms.

This article is part of the discussion meeting issue ‘Extracellular vesicles and the tumour microenvironment’.

1. Introduction

The tumour microenvironment (TME) plays an essential role in cancer development, metastasis and drug resistance [1]. Cancer-associated fibroblasts (CAFs) are considered as a major component of the TME [2–5]. However, the origin of CAFs is still controversial [2,6]. Relative to solid tumours, the bone marrow microenvironment (BME), where haematopoietic stem cells (HSCs) are located, is considered much more complicated (figure 1). First, HSCs reside in a specialized microenvironment called the haematopoietic niche [7]. The endosteal niche, hallmarked by osteoblasts lining the bone marrow cavity, has been proposed as an important source of HSC quiescence (figure 1, top right), while the vascular niche has been implicated in HSC maintenance and regeneration (figure 1, bottom right). Second, the definition of CAFs is not generally used to indicate the tumour-surrounding cell population in haematological neoplasms. Such cells are referred to as bone marrow mesenchymal stromal cells (BM-MSCs) [8]. While BM-MSCs and CAFs may share biological properties, especially in multiple myeloma (MM) [9] and chronic lymphocytic leukaemia (CLL) [10], the differences and similarities between BM-MSCs and CAFs are not fully elucidated in leukaemias [8]. In addition to BM-MSCs, immune cells, such as T cells, B cells and natural killer (NK) cells, also interact with tumour cells. Myeloid-derived suppressor cells (MDSCs) represent another population of regulatory cells, which impair anti-tumour innate and adaptive immune responses [11].

Figure 1.

Bone marrow microenvironment in haematological neoplasms. The bone marrow microenvironment is located in bone. Two types of niches, the endosteal niche and vascular niche, are located adjacent to the bone cortex. Tumour cells secrete extracellular vesicles (small green circles) towards their surrounding cells such as bone marrow mesenchymal stromal cells (BM-MSCs), endothelial cells and immune cells. Various types of cell–cell communication are assumed, such as tumour cells to BM-MSCs, tumour cells to endothelial cells, tumour cells to immune cells, and others. MDSC, myeloid-derived suppressor cell; NK, natural killer; RBC, red blood cell.

It is believed that cell–cell communication between tumour cells and their surrounding cells is modulated by direct cell contact and soluble factors such as cytokines. Emerging evidence suggests that extracellular vesicles (EVs) shed from either tumour cells or their surrounding cells act as a mediator of crosstalk in the TME [12] because EVs contain complex cargo including proteins, mRNAs, lipids and microRNAs (miRNAs). The current generally recognized categories of EVs are exosomes (the most well-known category) with a size of 30–100 nm, microvesicles with a size of 100–1000 nm, and large oncosomes ranging in size from 1 to 10 µm [13,14]. To date, most EV studies in haematologic neoplasms refer to exosomes or microvesicles. However, a recent study by Johnson et al. focused on the heterogeneity of EVs and characterized large EVs in acute lymphoblastic leukaemia [15]. In this review, we highlight and discuss recent studies of EV-mediated cell–cell communication in haematological neoplasms, particularly leukaemia and MM.

2. Acute myeloid leukaemia-derived extracellular vesicles in the bone marrow microenvironment

Leukaemia is a type of haematological neoplasm which arises from HSCs. Leukaemia types are subdivided by cell lineage (myeloid or lymphoid) and clinical course (acute or chronic). High throughput genomic analyses have revealed that genetic alteration of leukaemic cells is likely a more important determinant of disease severity [16]. However, increasing evidence suggests that leukaemic cell-derived EVs affect their surrounding cells in autocrine and/or paracrine manners [17–19].

In acute myeloid leukaemia (AML), Kumar et al. found evidence that AML-EVs alter the BME to facilitate leukaemic cell growth and suppress normal haematopoiesis in a mouse model [20]. The unique miRNA profile of AML-EVs, including miR-155, has the potential to increase leukaemic fitness by dysregulation of other cell types in the BME [21]. Myelodysplastic syndrome (MDS) is a clonal myeloid neoplasm characterized by ineffective haematopoiesis, and approximately 30% of patients develop AML. In addition to AML-EVs, MDS-EVs are linked to stromal cell dysfunction. Therefore, EV-mediated cell–cell interaction is also involved in bone marrow failure syndrome [22]. Although various components of AML-EVs, such as proteins, mRNAs and miRNAs, have been identified [17,21], it is difficult to focus on a single pathway for AML. Another important issue is that AML-EVs suppress immune cells such as NK cells [18]. In serum obtained from AML patients, AML-microvesicles mediate suppression of NK cell activity via the transforming growth factor-β1 signalling pathway, and interleukin (IL)-15 protects NK cells from the adverse effects of AML-microvesicles [18].

To date, there is no appropriate in vitro model to elucidate the complex cell–cell interactions in the bone marrow niche where leukaemic stem cells exist. Crosstalk via EVs between osteoblasts and HSCs or between BM-MSCs and HSCs also remains to be resolved. For this reason, much effort has been made to determine the diagnostic value of circulating EVs rather than the mechanism of cell–cell communication in AML [23]. A recent report by Viola et al. demonstrated that EVs derived from BM-MSCs induce tyrosine kinase inhibitor resistance in AML [24], suggesting a new therapeutic approach targeting BM-MSCs in AML. Considering the critical roles of EVs in AML, understanding the mechanisms regulating signalling pathways in recipient cells may provide additional insights into the use of EVs as therapeutic agents for treating AML.

3. Bone marrow angiogenesis and extracellular vesicles derived from chronic myeloid leukaemia

Chronic myelogenous leukaemia (CML) characterized by the BCR-ABL chimeric protein is another type of myeloid leukaemia, which tends to progress more slowly than AML. Increased microvessel density and clinicopathological correlations with bone marrow angiogenesis have been reported in CML patients [25]. We and others have shown that EVs secreted by CML cells can potentially influence in vitro and/or in vivo angiogenesis by directly affecting the properties of endothelial cells [26,27]. Taverna and colleagues first provided direct evidence that fluorescent-labelled EVs released by K562 cells are internalized by human umbilical vein endothelial cells (HUVECs) during tubular differentiation on Matrigel, thereby enhancing angiogenesis [26,28]. They also demonstrated that functional transfer of CML-EV-miR126 targets C-X-C motif chemokine ligand 12 and vascular cell adhesion molecule in HUVECs [29]. These findings indicate that exogenous miRNAs transferred via EVs function similarly to endogenous miRNAs in HUVECs. CML-EVs also induce increased secretion of IL-8 in BM-MSCs, thereby promoting leukaemic cell growth in vitro and in vivo [30].

Because hypoxia is known to be a regulator of angiogenesis, we investigated how hypoxia triggers EV-mediated angiogenesis using the human leukaemic cell line K562. We found that K562 cell-EVs under hypoxic conditions (1% O₂ for 24 h) significantly enhance tube formation of HUVECs compared with EVs produced under normoxic conditions (20% O₂ for 24 h) [31] (figure 2). These experiments employed artificial in vitro conditions for short-term exposure to hypoxia. However, we found that miR-210 in EVs (EV-miR-210) downregulates ephrin A3 in HUVECs and alters EV components under hypoxic conditions, thereby affecting the behaviour of HUVECs [31]. EVs shed from leukaemic cells act in the following manners. First, leukaemia-EVs educate BM-MSCs for their own cell survival. As a result, normal haematopoiesis is suppressed. Second, leukaemia-EVs induce immune suppression by targeting T cells, B cells and NK cells. Third, leukaemia-EVs induce angiogenesis, especially in CML.

Figure 2.

Hypoxia and angiogenesis in leukaemia. Human umbilical vein cells (HUVECs) were cultured with or without extracellular vesicles (EVs) secreted from a human leukaemia cell line (K562). The cells were cultured for 24 h and then EVs were collected. In normoxia, tube formation of HUVECs was increased in the presence of K562-derived EVs compared with control HUVECs (middle). When EVs obtained from K562 cells cultured under hypoxic conditions (K562-EV-hypoxia) were added to HUVECs, tube formation was much more prominent (bottom). These findings indicate possible roles of EVs, hypoxia and angiogenesis in leukaemia.

4. Extracellular vesicle-mediated cell–cell communication in multiple myeloma

MM is a blood cancer formed by malignant plasma cells that create tumour masses in bone. In contrast to leukaemia, the tumour masses are predominantly located in the bone marrow. Therefore, cell–cell communication in the BME has a key role in the pathogenesis of MM. Because targeting MM cells alone is insufficient to establish curable treatment strategies for MM, the interaction between MM cells and their surrounding cells has been studied extensively [32]. In terms of clinical practice, circulating EV-miRNAs from newly diagnosed MM patients are a potential biomarker for prognosis [33]. Because MM appears to be pathophysiologically similar to solid tumours, a cell-free specimen has great benefit for evaluating minimal residual disease. We discuss the characteristic BME in MM, especially hypoxia, in the following section. We then summarize the role of MM-EVs, BM-MSC-EVs and the mechanism of EV transfer in MM.

Bone marrow is hypoxic in nature, and oxygen tension in MM-infiltrated BM is lower than that in normal BM [34]. The massive proliferation of MM cells produces hypoxic conditions in the tumour, which may lead to more rapid cell growth, drug resistance and angiogenesis [35]. Therefore, we focused on the effect of hypoxia in the BME and established three MM cell lines that could be adapted to long-lasting hypoxia, namely hypoxia-resistant MM cells (HR-MM cells). Mutation analysis of HR-MM cells revealed no acquired genetic variation compared with their parental cells, while the gene expression profile of HR-MM cells was modified during the adaptation process [36]. We considered that HR-MM cells may be suitable for analyses of the BME, which mimics the in vivo state because most in vitro hypoxia studies use short-term hypoxic exposure (3–24 h) which may reflect acute hypoxic shock rather than the MM-BME. Although no significant difference existed in the size of EVs between HR-MM and parental cells, the amount of EVs released from HR-MM cells was approximately twofold greater. We also found that HR-MM cells enhanced in vitro tube formation and increased angiogenesis in vivo in a Matrigel assay. It was notable that EV-miR-135b from HR-MM cells enhanced angiogenesis by direct suppression of factor-inhibiting hypoxia-inducible factor 1 (FIH-1) in endothelial cells. It remains uncertain how other components within HR-MM cell-EVs, such as proteins, mRNAs and miRNAs, may be involved in EV-mediated angiogenesis.

Regarding the cargo of MM-EVs, three independent proteomic studies revealed the significance of proteins in EV-mediated intercellular communication [37–39]. Harshman et al. identified 583 MM-EV proteins in total by liquid chromatography coupled with tandem mass spectrometry [37]. They found that MM-EVs contained molecules related to a distinct pathway, such as antigen-presenting molecules (major histocompatibility complex class I and II) and adhesion molecules (tetraspanins and integrins), in MM cell lines MM.1S and U266 [37]. MM-EVs contain the biologically active form of CD147, which drives MM cell proliferation [38]. In addition, MM-EVs harbour CD138, which is known to be an angiogenic regulator [39], and the angiogenic activity has been confirmed using mouse MM-EVs in vivo [40].

MM-EVs also interact with BM-MSCs to establish a favourable BME for MM: they promote an increase of miR-146a in BM-MSCs, which induces more cytokines and chemokines, such as IL-6, and support MM cell growth [41] (figure 3). Another example of tumour cell and BM-MSC cooperation has been reported in another type of lymphoid malignancy, CLL. CLL-derived EVs are internalized by BM-MSCs and deliver functionally active components to BM-MSCs [10]. Taken together, tumour-derived EVs have the ability to transform BM-MSCs to CAF-like cells. Of note is that MM-EVs induce prolonged survival of MDSCs which have a strong immune-suppressive activity in cancer patients [40]. Raimondi et al. demonstrated the effect of MM-EVs on both human primary osteoclasts and murine pre-osteoclastic RAW264.7 cells [42]. These results indicate that MM-EVs are key regulators that educate BME-targeting endothelial cells, BM-MSCs, MDSCs and osteoclasts.

Figure 3.

Cell–cell communication by extracellular vesicles in multiple myeloma. Massive proliferation of abnormal plasma cells is shown in the centre. The centre of the mass has pale blue plasma cells, indicating the hypoxic state. In addition to conventional signalling pathways, such as direct cell adhesion and soluble factors, either extracellular vesicles (EVs) derived from multiple myeloma (MM) cells (MM-EVs; small green circles) or EVs derived from bone marrow mesenchymal stromal cells (BM-MSC-EVs; small orange circles) modify the behaviour of recipient cells as follows. MM-EVs modulate BM-MSCs and induce a favourable microenvironment for MM cells. Conversely, BM-MSC-EVs also induce MM cell growth, survival and drug resistance (bottom right). MM-EVs induce angiogenesis. In a long-lasting hypoxic state, EV-miR-135b acts as a mediator of angiogenesis through the FIH-1/hypoxia-inducible factor-1 signalling pathway (bottom left). BM-MSC-EVs interact with myeloid-derived suppressor cells (MDSCs), and MM-EVs modulate immune cells, such as natural killer (NK) cells, resulting in immunosuppression (top right). MM-EVs activate osteoclasts which may be linked to osteolysis in MM (top left).

In addition to MM-EVs, EVs derived from BM-MSCs facilitate MM cell growth by maintaining a favourable BME for MM. Roccaro et al. reported the difference in EV-miRNA contents between MM-BM-MSCs and normal BM-MSCs [43]. EVs derived from MM-BM-MSCs have less tumour suppressor miR-15a and higher levels of oncogenic proteins, cytokines and adhesion molecules, whereas normal-BM-MSCs have the potential to inhibit MM cell growth. Their findings indicate a CAF-like ability of MM-BM-MSCs and the possible use of normal BM-MSCs for MM cell therapy. We also found that BM-MSCs from healthy young donors inhibit MM angiogenesis through EV-miRNA [44]. From the aspect of CAFs, BM-MSC-derived EVs induce proliferation, survival and drug resistance of MM cells [45], and they are able to activate MDSCs [46]. As a result, BM-MSC-derived EVs contribute to not only the BME supporting MM cell growth but also immunosuppression related to disease progression.

Recent studies have focused on the mechanisms of EV release and uptake in MM. EV-fibronectin has recently been identified as a key binding ligand for heparan sulfate that can act as a receptor for EV uptake by MM cells [41]. Conversely, Bcl-xL is an exosomal caspase-3 substrate, and its processing is required for the uptake of exosomes by recipient cells [42]. In solid tumours, tumour-derived EVs use tropism to transfer into organ-specific recipient cells, and this tropism depends on distinct integrin expression patterns [43]. The mechanism by which EVs are captured and why EVs can select recipient cells remain to be elucidated in MM. Although the BME in MM is controlled by either a cytokine/chemokine network or cell adhesion molecules [32], interactions between conventional and EV-mediated pathways are much more complex than ever imagined.

5. Perspectives

Despite the fact that most studies have focused on tumour cell- and BM-MSC-derived EVs, how EVs derived from other types of cells act in the BME is largely unknown. Because a category of EVs includes components other than exosomes, including oncosomes, the type of EV and the cells from which EVs originate should be clarified to elucidate the precise mechanism of EV-mediated cell–cell communication. We identified EV-miRNA as an angiogenic factor in our experimental model. However, analysis of plasma EV-miR-135b did not simply reflect MM disease severity. It is likely that EV-miRNA in the proximal area (i.e. bone marrow) is not linked to EV-miRNA in remote areas such as circulating blood. The manner of cell–cell communication in the proximal area might be different from that in the distal area. Therefore, careful attention is needed for diagnostic use of EV-miRNA. To better understand EV-mediated cell–cell communication in haematological neoplasms, the origin of EVs, their tropism, and mechanism of EV transfer are emerging issues that need to be addressed.

Data accessibility

This article has no additional data.

Authors' contributions

T.U. performed part of the study discussed in this review. J.H.O. drafted the manuscript. K.O. helped to draft the manuscript. All authors read and approved the final manuscript.

Competing interests

We have no competing interests.

Funding

This review was supported by the Private University Strategic Research-Based Support Project (grant no. S1311016) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.