Transforming the Niche: The Emerging Role of Extracellular Vesicles in Acute Myeloid Leukaemia Progression

Abstract

Acute myeloid leukaemia (AML) management remains a significant challenge in oncology due to its low survival rates and high post-treatment relapse rates, mainly attributed to treatment-resistant leukaemic stem cells (LSCs) residing in bone marrow (BM) niches. This review offers an in-depth analysis of AML progression, highlighting the pivotal role of extracellular vesicles (EVs) in the dynamic remodelling of BM niche intercellular communication. We explore recent advancements elucidating the mechanisms through which EVs facilitate complex crosstalk, effectively promoting AML hallmarks and drug resistance. Adopting a temporal view, we chart the evolving landscape of EV-mediated interactions within the AML niche, underscoring the transformative potential of these insights for therapeutic intervention. Furthermore, the review discusses the emerging understanding of endothelial cell subsets' impact across BM niches in shaping AML disease progression, adding another layer of complexity to the disease progression and treatment resistance. We highlight the potential of cutting-edge methodologies, such as organ-on-chip (OoC) and single-EV analysis technologies, to provide unprecedented insights into AML-niche interactions in a human setting. Leveraging accumulated insights into AML EV signalling to reconfigure BM niches and pioneer novel approaches to decipher the EV signalling networks that fuel AML within the human context could revolutionise the development of niche-targeted therapy for leukaemia eradication.

Keywords: acute myeloid leukaemia; bone marrow microenvironment; exosomes; extracellular vesicles; organ-on-a-chip.

Figure 1

Landscape of extracellular particle (EP) and extracellular vesicle (EV) subtypes. Biogenesis and key characteristics of various EPs including both non-vesicular extracellular particles and a plethora of distinct EV subtypes. EVs are membrane-enclosed structures shed by all cell types, with critical roles in cell–cell communication and cargo delivery. Notably, leukaemic blasts can exploit EVs to reprogram marrow niche cells, promoting their own survival, proliferation, dormancy, and ultimately contributing to therapy resistance and relapse. (A) Non-vesicular extracellular particles (NVEPs) encompass a heterogeneous group of nanoparticles distinct from EVs. Unlike EVs, they lack a lipid bilayer membrane and are formed through various non-vesicular pathways. Examples include exomeres, shed from the plasma membrane, and supermeres, formed by protein aggregation. Symbol: (?) denotes the ill-defined nature of non-vesicular extracellular particles with largely unknown biogenesis and mechanisms of action. (B) Exosomes are the smallest EV subtype, ranging from 30 to 120 nm in diameter. They originate from the endosomal system. Invaginations of the limiting membrane of multivesicular bodies (MVBs) create intraluminal vesicles (ILVs) that become exosomes upon MVB fusion with the plasma membrane. Rab GTPases play a crucial role in directing MVB trafficking and exosome secretion, along with other essential molecules like SNARE proteins. (C) Ectosome/microvesicles are larger than exosomes, with a size range of 80–500 nm. Ectosomes bud directly outward from the plasma membrane in a Rho A- and ARF6-dependent process. Increased calcium concentration and cortical actin assembly at the budding site facilitate their formation. (D) Migrasomes are large EVs, ranging from 500 to 3000 nm. Migrasomes contain smaller EVs within their lumen and originate from the fragmentation of retraction fibres formed during cell migration. TSPAN4 proteins are crucial for migrasome formation, and damaged mitochondria are often found within them. (E) Oncosomes are a specialized type of ectosome released by cancer cells. Their biogenesis is highly heterogeneous and cancer-type-dependent, reflecting the diverse mechanisms employed by different cancers to manipulate their environment. They contribute to multiple hallmarks of cancer progression. (F) Apoptotic bodies, remnants of programmed cell death (apoptosis), are the largest EVs, ranging from 100 to 5000 nm. These vesicles form from the fragmentation of the apoptotic cell and are subsequently released. Caspase-3 and ROCK1 are key players in the apoptotic process that leads to the formation and release of apoptotic bodies.

Figure 2

Genetic and/or epigenetic events driving the inception of AML. (A) Stepwise progression from healthy HSCs to LSCs. Initially, HSCs can accumulate the first genetic or epigenetic “hit”, leading to the formation of pre-leukaemic stem cells (pre-LSCs). These pre-LSCs expand within the BM environment, carrying mutations in tumour suppressor genes (e.g., TP53) and epigenetic regulators (e.g., DNMT3A, ASXL1, IDH1/2, and TET2). Additional mutations (the “second hit”) occur in pre-LSCs, affecting genes that control HSC quiescence, proliferation, self-renewal, differentiation, or epigenetic regulation. Accumulation of these two sets of mutations results in the emergence of full-blown AML. LSCs can also arise from pre-leukaemic multipotent progenitors (MPPs) that acquire a second genetic or epigenetic alteration concomitantly with a third event to regain HSCs’ self-renewal ability. (B) Upon these genetic events, the successful establishment of LSCs relies on their autocrine signalling ability to survive. This early survival is reliant mostly on autocrine EV signalling (alongside with HSCs and pre-LSC counterparts) that shed exosomes enriched in stemness factors such as TPO, ANGPTL2/3, and other EVs carrying pro-leukaemic miRs that upregulate LSCs’ Wnt, NFkB, and/or PI3K/AKT pathways, driving unchecked AML blast cell proliferation. EV shedding is exacerbated in transformed LSCs compared to healthy HSC counterparts, highlighting the importance of this mechanism for the onset of AML. Autocrine EV endorsement of multiple leukaemic cancer hallmarks is further complemented by autocrine secretion of several soluble cytokines. Arguably, many of those cytokines can also be caried in the EV luminal compartment. Symbols: (*) represents the “first hit” early mutations occurring in HSCs/MPPs that give rise to pre-LSCs; (**) illustrates the “second hit” mutations that pre-LSCs accumulate in order to transform in LSCs; in the case of pre-transformed MPPs (***) denotes the existence of a “third event” in genes that enables MPPs to step back and re-gain the unlimited self-reviewal ability of HSCs. (+) Somatic alterations in HSCs give rise to an expanded pool of MPPs carrying at least one AML driver mutation facilitating the acquisition of additional alterations required for AML establishment.

Figure 3

Mapping EV signalling networks driving AML. Blueprints of BM niche reprograming in distinct stages of BM infiltration. Healthy BM is organized in two functionally distinct compartments: a stiffer endosteal niche at the inner surface of the bone and the vascular niche located at the core of marrows’ cavity. Both haematopoietic niches are required for HSCs’ maintenance, proliferation, and maturation along lymphoid and myeloid lineages. AML onset severely remodels the BM microenvironment, causing (i) abnormal accumulation of osteoprogenitors and the (ii) destruction of the lining endosteum and endosteal vasculature. This leads to (iii) the expansion of central cavity space available for AML proliferation supported by an abnormal-vascularization BM core. Drastic remodelling of haematopoietic niches drives AML blasts to outcompete healthy HSCs and progenitors, compromising normal haematopoiesis. (A) AML blasts shed high amounts of EVs carrying pro-inflammatory and osteogenic factors, including BMP-2, miR-4532, or YBX1, that drive BMMSCs to expand and differentiate into osteoprogenitors. However, several of these EVs increase DKK1 expression at osteoprogenitors, decreasing the Wnt signalling required for the osteoblast maturation step. Osteolineage differentiation stalling causes abnormal accumulation of osteoprogenitors, which typically provide signalling molecules required for HSC maintenance, activation, and proliferation. (B) On the other hand, mature osteoblasts—that seem to be more involved in providing support for HSC maturation and differentiation—are typically destroyed by the inflammatory environment throughout disease progression. Early on, osteoblasts seem to be protective against AML. AML blasts balance this roadblock by exploiting the kynurenine(kyn)–HTR1B–SAA–IDO1 axis by secreting exacerbated amounts of kynurenine that interact with mature osteoblast receptor HTR1B, triggering the release of SAA that in turn activates the AHR pathway in AML blasts with transcription of IDO1 and other molecules that stimulate AML proliferation. With increasing clone expansion and inflammatory-driven cell death of lining osteoblasts over time, the kyn-SAA positive feedback loop eventually surpasses the inhibitory secretion profile of osteoblasts towards overt disease. (C) During AML establishment, LSCs seem to secrete EVs carrying L-plastin, PDRX2/4, and EPO that transiently increase the number of osteoclasts degrading the endosteal bone. This may assist the thriving of AML to overt disease by expanding the available marrow cavity space for clonal expansion. During disease progression, osteoclast numbers decrease, yet their activity may still be endorsed by high EPO levels secreted by expanded AML clones and by erythroid progenitor cells (EPCs) facing secondary anaemia stimuli triggered by haemopoiesis disruption by high AML infiltration. (D) LSCs secrete EVs carrying pro-angiogenic factors such as VEGF, VEGFR, IL-8, miR92-a, or IGF1R mRNA towards central niche vasculature. These EVs seem to target both perivascular MSCs and pericytes (that in return secrete high amounts of VEGF) and endothelial cells. The exacerbated pro-angiogenic microenvironment triggers an abnormal angiogenic process that drives the increased microvessel density with the formation of disorganized and leaky microcapillaries. In return, endothelial cells shed ANGPL2 + EVs that interact with LILRB2/4 receptors at AML blasts to reinforce their proliferation. The exposure of the remaining healthy HSCs to high reactive oxygen species (ROS) levels triggers their displacement from the niche with a possible exhaustion of the pool of quiescent HSCs. On the other hand, a residual reservoir of LSCs is guaranteed through the incorporation of AML blasts into the vasculature as quiescent tissue-associated AML cells. Their remarkable plasticity ensures a steady reservoir of residual disease. (E) Arguably, AML blasts shedding EVs seem to trigger certain MSC subsets to differentiate into adipocytes, losing their ability to support healthy HSCs. Subsequently, AML reprograms these adipocytes via GDF15, yielding smaller adipocytes that release fatty acids (FAs) into the vascular milieu. Upregulated FABP4 in nearby AML blasts transports these FAs to the mitochondria, generating the energy needed for overt leukaemia growth. In return, adipocytes secrete factors that endorse AML blast proliferation and resistance of cell death. (F) Sympathetic neurons are typically destroyed in the late stages of blast infiltration, originating generalised marrow neuropathy. The mechanisms of neuronal targeting in AML remains largely unexplored. Some reports observed in other haematological malignancies that neurons secrete atypical EVs in therapy-refractory patients in childhood acute lymphoblastic leukaemia (ALL). Symbols: (X) elements of the bone marrow niche that are frequently obliterated in AML; (?) denotes the largely unexplored nature of EV-mediated communication in AML disease.