Understanding the interaction between leukaemia stem cells and their microenvironment to improve therapeutic approaches

Abstract

Although chemotherapeutic regimens can eliminate blasts in leukaemia patients, such therapies are associated with toxicity and often fail to eliminate all malignant cells resulting in disease relapse. Disease relapse has been attributed to the persistence of leukaemia cells in the bone marrow (BM) with the capacity to recapitulate disease; these cells are often referred to as leukaemia stem cells (LSCs). Although LSCs have distinct characteristics in terms of pathobiology and immunophenotype, they are still regulated by their interactions with the surrounding microenvironment. Thus, understanding the interaction between LSCs and their microenvironment is critical to identify effective therapies. To this end, there are numerous efforts to develop models to study such interactions. In this review, we will focus on the reciprocal interactions between LSCs and their milieu in the BM. Furthermore, we will highlight relevant therapies targeting these interactions and discuss some of the promising in vitro models designed to mimic such relationship.

LINKED ARTICLES:

This article is part of a themed issue on Cancer Microenvironment and Pharmacological Interventions. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v181.2/issuetoc

1 |. INTRODUCTION

Stem cells are typically defined by their self-renewal, differentiation and proliferative capacities (Guzman & Jordan, 2004; Reya et al., 2001). In the context of normal mammalian haematopoiesis, haematopoietic stem cells (HSCs) have the capacity to replenish the entire blood cell system (Bonnet, 2002; Morrison et al., 1995). Extrinsic signals from the microenvironment in bone marrow (BM) are key for haematopoietic cell fate decisions (Comazzetto et al., 2021). Leukaemia can originate from the acquisition of oncogenic events in haematopoietic stem/progenitor cells (HSPCs) (Bonnet & Dick, 1997), giving rise to leukaemia stem cells (LSCs). LSCs, as HSCs, obtain cues from the BM microenvironment for their survival, self-renewal and proliferation (Tabe & Konopleva, 2014, 2015). Therefore, taking into account the interactions between LSCs and the surrounding microenvironment may result in the improvement of therapeutic interventions to eradicate LSCs.

2 |. LEUKAEMIA STEM CELLS ARE THE RESERVOIR OF THE DISEASE

LSCs, also known as leukaemia initiating cells (LICs), were initially identified circa 1994 (Lapidot et al., 1994) by showing that samples obtained from acute myeloid leukaemia (AML) patients can be serially transplanted into immunodeficient mice. Importantly, such cells share, in part, the immunophenotype of HSCs (Lapidot et al., 1994). LSCs, as HSCs, are organized in a hierarchical manner where they reside at the top being capable giving rise to, in this case, malignant progeny (Bonnet & Dick, 1997; Felipe Rico et al., 2013; Roboz & Guzman, 2009). Furthermore, like their normal counterparts, LSCs have the capacity to self-renew and proliferate (Bonnet, 2002; Bonnet & Dick, 1997; Felipe Rico et al., 2013; Laurenti & Gottgens, 2018; Mojtahedi et al., 2021; Morrison et al., 1995; Orkin & Zon, 2008; Roboz & Guzman, 2009). In addition to AML, malignant stem cells have been described in other haematologic malignancies including myelodysplastic syndromes (MDS), chronic myeloid leukaemia (CML) and acute lymphoblastic leukaemia (ALL) (Cox et al., 2009; Ebinger et al., 2016; Joudinaud & Boyer, 2021; Vetrie et al., 2020).

Increasing evidence indicates that the LSC frequency at diagnosis in AML patients is of prognostic relevance (Kamel et al., 2022; Terwijn et al., 2014; van Rhenen et al., 2005). The persistence of LSCs after treatment has been shown when assessing measurable residual disease (MRD) in AML patients (Gomez-Arteaga & Guzman, 2018; Zeijlemaker et al., 2016). The evaluation of MRD constitutes the most important indicator of response to therapy in all types of acute leukaemia, which helps in the selection of therapy including the decision of stem cell transplantation (Gomez-Arteaga & Guzman, 2018; Kamel et al., 2022).

Despite the biological differences in LSCs among leukaemia patients (Vetrie et al., 2020), their survival and fate are tightly regulated by interactions with their niche. Such interactions can result in protection from chemotherapy contributing to leukaemia relapse (Felipe Rico et al., 2013; Mojtahedi et al., 2021).

3 |. BONE MARROW NICHE

The BM is a soft viscous tissue that occupies cavities in the bone. HSCs require the interactions with neighbouring cells that constitute the BM microenvironment, also termed “the niche”. In haematology, the use of niche was first utilized by Schofield R in 1978 (Schofield, 1978), who proposed that stem cells were localized in specialized sites that dictate their fate (self-renew or differentiate). Soon after, understanding the components of the niche that can control cell fate became of scientific interest and led to the definition of osteogenic and haematopoietic niches (Bianco, 2011b; Mendelson & Frenette, 2014).

The BM niche is composed of different cell types, soluble factors and extracellular matrix (ECM). Two different niches have been anatomically described within BM: The endosteal and vascular niches (Kiel & Morrison, 2008) (Figure 1). Arterioles run close to the endosteal surface that contains bone-lining cells (osteoblasts and osteoclasts), sympathetic nerve fibres unsheathed by non-myelinating Schwann cells, and immune cells such as regulatory T (Treg) cells (Fujisaki et al., 2011; Kunisaki et al., 2013). This endosteal niche maintains HSCs mainly in a quiescent state (Kunisaki et al., 2013). Fenestrated sinusoids transport less-quiescent HSCs to the vascular niche (Kunisaki et al., 2013). Mesenchymal stem/stromal cells (MSCs) are found in the perivascular area and in the endosteal niche (Bianco, 2011a; Ehninger & Trumpp, 2011; Kunisaki et al., 2013; Mendez-Ferrer et al., 2010). In addition to cellular components, the ECM plays a role in regulating quiescence of HSCs in both niches. Specifically, endosteal niche-containing fibronectin is stiff and supports quiescence in HSCs, whereas vascular niche-containing laminin is soft and supports their proliferation (O’Reilly, Zeinabad, & Szegezdi, 2021). MSCs are thought to be the dynamic organizers of the BM niche, regulating both bone homeostasis and blood flow. Because of their capacity to differentiate into osteoblasts and regulate osteoclasts, MSCs maintain the balance between osteogenesis and osteoclastogenesis (Fernandez Vallone, Hofer, et al., 2013; Fernandez Vallone, Romaniuk, et al., 2013; Sanmartin et al., 2021). MSCs also are able to differentiate to adipocytes (Fernandez Vallone, Hofer, et al., 2013; Fernandez Vallone, Romaniuk, et al., 2013; Sanmartin et al., 2021). Adipogenesis and lipolysis at the sinusoid wall in BM regulate the niche space and blood circulation (Bianco, 2011a).

FIGURE 1.

Simplified illustration of the bone marrow microenvironment in leukaemia. The niche includes different cell types such as osteoblasts, osteocytes, osteoclasts, nerve fibres, regulatory T (Treg) cells, NK cells, CD8⁺ T cells, endothelial cells, mesenchymal stem/stromal cells (MSCs) and adipocytes among others. Key molecules involved in the traffic of haematopoietic stem cells (HSCs) and leukaemia stem cells (LSCs) are included. The lightning bolt represents a transforming event that affects the osteoblast function represented in grey. Treg cells are necessary for HSC persistence. LSCs also can alter the action of CD8⁺ T cells and NK cells. CXCL12: C-X-C motif chemokine 12. CXCR4: C-X-C chemokine receptor type 4. ECM: extracellular matrix. FAO: fatty acid oxidation. FFA: free fatty acids. VCAM-1: vascular cell adhesion molecule 1. VLA-4: very late antigen-4.

In leukaemia, LSCs have been found to interact with endothelial cells, osteoblasts, MSCs, adipocytes, sympathetic neural cells and other immune cells by direct contact or soluble mediators in the BM (Duhrsen & Hossfeld, 1996; Tabe & Konopleva, 2017).

4 |. LEUKAEMIA STEM CELLS AND ENDOTHELIAL CELLS

The main function of endothelial cells is to regulate the migration of haematopoietic cells between the BM and peripheral blood. Vascular cell adhesion molecule 1 (VCAM-1) and E-selectin (CD62E) are the major inducible endothelial cell adhesion molecules that play a role in the migration of both normal HSCs and LSCs (Cavenagh et al., 1993; Godavarthy et al., 2020).

In AML, the vascular niches in the BM can be remodelled by factors produced by leukaemic cells, reducing the capacity to support HSCs (Duarte et al., 2018; Passaro et al., 2017). Using small molecules that can prevent such remodelling has been shown to improve response to chemotherapy in AML animal models. In an fms-like tyrosine kinase 3-internal tandem duplication (FLT3-ITD) AML mouse model, the BM niche showed a loss of arterioles and a gain of sinusoids driven by tumour necrosis factor shed form (TNFα) released by AML blasts. TNFα downregulates microRNA (miR)-126 in endothelial cells that are lining the arterioles. Finally, treatment of FLT3-ITD AML with tyrosine kinase inhibitors (TKIs) decreased TNFα levels due to a cytoreduction of blasts, leading to an increased expression of miR-126 in endothelial cells and resulting in protection and persistence of LSCs (Zhang et al., 2021).

Taken together, therapies that can interfere with vascular niche/LSCs interactions may offer a path to prevent chemoresistance driven by this niche. Some examples include E-selectin inhibitors, deferoxa-mine, and miR-126 deprivation.

5 |. LEUKAEMIA STEM CELLS AND OSTEOBLASTS

Osteoblasts are mononucleate cuboidal cells involved in bone formation and directly interact with other cells, such as HSCs and osteocytes. Osteoblasts secrete components of the ECM in the bone and can induce quiescence in HSCs and LSCs (Adams et al., 2006; Ishikawa et al., 2007), and may play a role in the oncogenic transformation of HSPCs into LSCs (Le et al., 2018). Aberrant osteoblasts can result in transforming events; for example, if they bear an activating mutation in β-catenin, they can induce AML in mice. Mechanistically, the interaction between forkhead box protein O1 (FoxO1) and β-catenin in osteoblasts increases the expression of Jagged-1, which induces the leukaemogenic transformation of HSCs by activation of Notch signalling (Kode et al., 2014, 2016). In a reciprocal way, leukaemia cells can modify osteoblasts to gain growth advantage. Using a transgenic CML mouse model (Schepers et al., 2013), it was demonstrated that leukaemic blasts alter transforming growth factor beta-1 (TGFβ1), Notch signalling, and inflammatory signalling in osteoblasts, to support LSCs. Moreover, malignant cells created an inhospitable microenvironment for normal HSCs, decreasing the expression of factors involved in HSC retention (Schepers et al., 2013).

Osteoblasts, among other cells, contribute to drug resistance in LSCs (Filik et al., 2021). A recent study indicated that the inhibition of phosphoinositide 3-kinase (PI3K) in osteoblasts was a new effective approach to fight the niche-induced TKI resistance in CML-stem cells (Filik et al., 2021).

6 |. LEUKAEMIA STEM CELLS AND MESENCHYMAL STEM/STROMAL CELLS

MSCs include a subset of self-renewing and multipotent progenitors with the unique capacity to organize the BM niche (Bianco, 2011a; Friedenstein et al., 1982). Considerable efforts have been made to define what BM-MSCs are. A pioneer in the field was Bianco and colleagues (Bianco & Robey, 2015), who have redefined BM-MSCs as skeletal stem cells localized in the BM sinusoids. Bianco and colleagues (Sacchetti et al., 2016) went further to demonstrate that transcriptomes for MSCs from different anatomical regions also are different. Among BM-MSCs, we can find C-X-C motif chemokine ligand (CXCL12)-abundant reticular (CAR) cells (Sugiyama et al., 2006), Nestin⁺ MSCs (Mendez-Ferrer et al., 2010) and leptin receptor (LepR)⁺ MSCs (Ding et al., 2012). All these stromal cells exhibit significant overlap in terms of the immunophenotype (Morrison & Scadden, 2014).

HSCs are in direct contact with CAR cells on the sinusoidal areas of the BM. Most CAR cells express peroxisome proliferator-activated receptor-γ (PPARγ), runt-related transcription factor 2 (Runx2) and Osterix, and are adipo-osteogenic bipotential progenitors (Omatsu et al., 2010). CAR cells act as professional cytokine-secreting cells, establishing peri-vascular micro-niches for maintenance and fate decisions of HSCs (Baccin et al., 2020). In addition to osteoblasts and Nestin⁺ MSCs, CAR cells produce high levels of CXCL12, also known as stromal cell-derived factor-1 (SDF-1) (Cordeiro Gomes et al., 2016). CXCL12 attracts HSCs expressing its cognate receptor [C-X-C chemokine receptor (CXCR4)] (Roversi et al., 2021). CXCL12-CXCR4 signalling is involved in homing of normal HSCs/HSPCs, as well as LSCs, into BM (Kim & Broxmeyer, 1998; Roversi et al., 2021). Therefore, aberrant expression of CXCR4 by LSCs or targeting CXCL12 affects the homing process (Jin et al., 2008). The specific deletion of CXCL12 in BM-MSCs increased the LSC numbers in a CML mouse model (P. Agarwal et al., 2019). In contrast, the specific deletion of CXCL12 in endothelial cells decreased LSC numbers (P. Agarwal et al., 2019), suggesting niche-specific effects. Clinical trials targeting CXCR4 have been performed in AML; however, despite the mobilization of LSCs observed in some patients, the clinical benefit of adding plerixafor, a CXCR4 antagonist, to AML treatment regimens remains uncertain (Roboz et al., 2018).

Depletion of Nestin⁺ MSCs significantly reduced BM homing of HSCs/HSPCs in vivo (Mendez-Ferrer et al., 2010). In the context of blood cancers, Nestin⁺ MSCs are consistently reduced in the BM of patients with preleukaemic disorders, such as myeloproliferative neoplasms (Arranz et al., 2014). In contrast, increased number of BM-Nestin⁺ MSCs have been reported in AML mouse models (Hanoun et al., 2014), contributing to survival and chemoresistance of LSCs (Forte et al., 2020).

Conditional deletion of stem cell factor (SCF) in LepR⁺ MSCs significantly reduced HSC number (Ding et al., 2012), whereas its deletion in osteoblasts did not affect HSC number and function (Ding et al., 2012). These findings emphasize the niche-specific functions. While LepR⁺ MSCs have been studied more extensively in normal haematopoiesis, the potential of these prominent niche cells to regulate LSCs is less understood.

7 |. LEUKAEMIA STEM CELLS AND ADIPOCYTES

Adipocytes are the dominant component of adult BM, and it has been suggested that they increase in number with ageing (Justesen et al., 2001). The adipocyte compartment works as a pool of HSCs and HSPCs (J. Han et al., 2010), because HSCs/HSPCs utilize fatty acid oxidation to generate energy to maintain their self-renewal capacity (Ito et al., 2012; Ito & Suda, 2014; Mistry et al., 2021).

Accumulated evidence suggests that obesity is associated with an increased risk of incidence of leukaemia and increased mortality (Castillo et al., 2012; Poynter et al., 2016). Recent work has shown a reciprocal interaction between adipocytes and leukaemic blast (Shafat et al., 2017) where primary AML blasts can induce lipolysis in vitro, allowing the traffic of free fatty acids (FFA) from adipocytes to AML blasts. AML cells use FFAs for fatty acid oxidation to provide energy to satisfy their metabolic demands (Maher et al., 2018; Yan et al., 2017). Using a mouse CML model, Ye et al. (2016) indicated that gonadal adiposity tissue (GAT) is an extramedullary reservoir of LSCs. They describe that GAT resident-LSCs have a pro-inflammatory profile characterized by high levels of interleukin-1α (IL-1α), 1L-1β and colony stimulating factor-2 (CSF2), which mediate lipolysis. GAT lipolysis resulted in an increase of fatty acid oxidation in LSCs, especially within in a sub-population expressing CD36. Finally, they demonstrated that GAT resident-CD36⁺ LSCs are resistant to conventional chemotherapy such as cytarabine.

Therefore, disrupting the fatty acid metabolism represents another avenue to potentially eradicate LSCs.

8 |. LEUKAEMIA STEM CELLS AND IMMUNE CELLS

In addition to the mesenchymal compartment, the leukaemic BM niche includes other cells, such as non-transformed immune cells, or the immune microenvironment.

T cells are dysregulated in presence of leukaemia. T-cell senescence and exhaustion are dominant aspects involved in the immune dysfunction of AML patients (Tang et al., 2020). In addition, AML patients have an immunosuppressive BM niche characterized by an accumulation of Treg cells (Shenghui et al., 2011). Treg cells express forkhead box P3 (Foxp3), and they represent a functional distinct subset of CD4⁺ T cells (Martinez et al., 2019). Treg cells can localize on the endosteal surface within BM and are necessary for HSPC persistence (Fujisaki et al., 2011). In the context of AML, Treg cells also contribute to leukaemia progression, because Treg cell ablation prolongs the survival of AML mice by promoting the anti-leukaemic cytotoxic effects of CD8⁺ T cells (Wang et al., 2020). In addition, AML cells express inducible T cell co-stimulator ligand (ICOSL) and can induce the expression of ICOS in Treg cells (Y. Han et al., 2018). These ICOS⁺ Treg cells produce IL-10, thus stimulating the proliferation of AML cells. Furthermore, disruption of ICOSL using a blocking antibody delayed leukaemia progression in mouse models, by preventing the induction of ICOS⁺ in Treg cells (Y. Han et al., 2018).

Macrophages are extremely sensitive to their microenvironment (Clark et al., 2020). They can acquire several intermediate phenotypes from classically activated macrophages (M1, anti-tumour) to alternatively activated macrophages (M2, pro-tumour) (Clark et al., 2020). In the BM, macrophages are located at the endosteum and in vascular regions and they help in maintaining and regulating HSC niches (Levesque et al., 2021). They also play an important role in the pathophysiology of haematological neoplasms and their treatment response (Levesque et al., 2021; Miari et al., 2021). A recent study showed that promoting macrophage-driven immunity against AML was a potential therapeutic avenue to increase the antibody-dependent phagocytosis of LSCs by macrophages (Alves da Silva et al., 2022). In addition, the humanized antibody Hu5F9-G4 (5F9; magrolimab) anti-CD47 is being testing in AML patients (NCT03248479, NCT04435691, NCT04778410). Because CD47, a do-not-eat-me signal, is overexpressed in LSCs to defend themselves against phagocytosis by macrophages (Melo Garcia & Barabé, 2021), Hu5F9-G4 treatment could avoid leukaemia relapse.

Myeloid malignancies also affect the cytotoxic action of CD8⁺ T cells and natural killer (NK) cells (Carlsten & Jaras, 2019; Knaus et al., 2018; Norde et al., 2011). For example, AML blasts can directly alter the viability, expansion and senescence of CD8⁺ T cells in vitro (Knaus et al., 2018). Furthermore, AML blasts, including LSCs, may evade the immune attack by decreasing the expression of ligands for activating receptors on NK (Paczulla et al., 2019). In this context, an Fc-engineered anti-CD123 antibody (CSL362) was developed to enhance the cytotoxic activities of NK cells (Busfield et al., 2014; Nievergall et al., 2014).

Although there is evidence of dysregulation in immune cell-subsets in leukaemia, the role of each of the populations within the immune microenvironment in regulating LSC homeostasis needs further investigation.

9 |. LEUKAEMIA STEM CELLS AND SYMPATHETIC NEURAL CELLS

The sympathetic nervous system has shown to play a role in driving HSC retention/release from the BM by regulating cells in the BM niche (Hoggatt & Pelus, 2011; Katayama et al., 2006). In the context of AML, using a mixed-lineage leukaemia (MLL)-AF9 mouse model, AML cells can remodel the BM niche by decreasing the number of sympathetic nervous system networks. Such changes result in a depletion of stromal cells critical for normal HSC function and thus favour leukaemic niches that facilitate the growth and maintenance of LSCs (Hanoun et al., 2014). Inhibition of β2 adrenoceptors increased the LSC pool (Hanoun et al., 2014). This study suggests that perturbation of the adrenergic system, as a therapeutic approach, may prevent the remodelling of the niche preserving HSC function and preventing leukaemogenesis.

10 |. LEUKAEMIA STEM CELLS AND MOLECULES INVOLVED IN THEIR BONE MARROW RETENTION

Like HSCs, LSCs are retained in specialized BM-niches promoting survival. The inhibition of these interactions represents therapeutic approaches to eradicate LSCs.

The CD44-hyaluronan (HA) axis is involved in the adhesion between AML cells and BM-stromal cells/ECM (Grenier et al., 2021; Gutjahr et al., 2021). Indeed, antibodies such as H90 inhibited the adhesion of CD44⁺ CD38⁻ AML blasts to HA, decreasing the leukaemia progression in a patient-derived xenograft (PDX) setting (Jin et al., 2006).

Among the β1 integrins, very late antigen-4 (VLA-4, or α4β1 integrin) on AML-LSCs interacts with fibronectin and VCAM-1 on stromal cells (Grenier et al., 2021). Thus, VLA-4-specific antibodies were able to improve the effect of cytarabine on the BM residual disease in an AML mouse model (Matsunaga et al., 2003). AS101, a tellurate compound that inhibits VLA-4, is currently under clinical evaluation.

CXCR4 antagonists can mobilize leukaemia cells from BM to circulation, which makes them more sensitive to chemotherapeutic agents (Su et al., 2021). However, new evidence indicated that chemotherapy-resistant leukaemia cells are not affected by CXCR4 antagonists (Su et al., 2021). This principle has been tested in a Phase I clinical trial, where we observed that plerixafor induced mobilization of AML-stem and progenitor cells in a subset of older patients, but the clinical benefit of plerixafor combined with decitabine was uncertain (Roboz et al., 2018). Observations suggest that other components of the niche should be considered and may be unique to each patient. Plerixafor has been combined with other agents and in combination with TKIs, but failed to decrease residual disease in CML mouse models (A. Agarwal et al., 2012).

The relevance of CXCL1-CXCR2 axis in CML was recently reported. TNFα induced the expression of CXCL1 in BM stromal progenitors, resulting in enhanced support of LSC function (P. Agarwal et al., 2021). Thus, CXCR2 inhibition could represent a therapeutic approach to target CML-stem cells in combination with other agents (P. Agarwal et al., 2021).

11 |. IMPLEMENTING IN VITRO MODELS TO STUDY LEUKAEMIA STEM CELL-NICHE INTERACTIONS

Typically, drug-testing methods include xenograft mouse models and co-cultures on a monolayer of stromal cells (2D cultures). Xenograft mouse models are not cost-effective and cannot successfully replicate the human BM niche because the cells are from two different species. 2D cultures do not take into consideration other factors, such as the hypoxia that occur in the BM niche, which play a role in maintaining LSCs and provide survival advantage properties.

Currently, there are many efforts to develop in vitro 3D systems to model the leukaemic niche. Examples of such systems include BM-MSCs trapped in scaffolds (Aljitawi et al., 2014; O’Reilly, Zeinabad, Nolan, et al., 2021), and scaffolds built with decellularized material (Li et al., 2018). These systems provide the opportunity to study leukaemic cell responses in the context of a microenvironment. Other in vitro 3D systems like hydrogels have been developed to mimic vascular niche (Bray et al., 2017), allowing the evaluation of leukaemia-vascular interactions. Other systems consist of microfluidic devices (Houshmand et al., 2017), where the system is perfused with culture medium to mimic the interstitial flow. The above-mentioned systems have been able to reproduce the drug resistance features observed by the BM microenvironment. Such systems are built with healthy stroma. Other 3D systems have been able to integrate different BM niches (Aleman et al., 2019; Ma et al., 2020). Recently, an organoid-like structure was implemented in a novel 3D co-culture system using BM-MSCs from ALL patients generating patient-derived leukaemic spheroids (PDLS). This co-culture system was shown to enrich and maintain LSCs from paediatric ALL patient samples ex vivo (Balandran et al., 2021). The 3D structures are formed by MSCs and create a hypoxic environment, and have the potential to be useful for tailored drug screening approach to find compounds to ablate LSCs (representing ex vivo avatars). We have begun to implement this model in myeloid leukaemia (Figure 2).

FIGURE 2.

(a) Schematic representation of patient-derived leukaemic spheroids (PDLS). Mononuclear cells (MNCs) are obtained by Ficoll gradient of primary bone marrow (BM) samples from leukaemia patients. A fraction of MNCs is used to generate mesenchymal stem/stromal cells (MSCs) to establish the spheroids later. Once spheroids are ready, another fraction of MNCs, which was previously cryopreserved, is used to generate the co-cultures. This approach allows for ex vivo leukaemia patient avatars to find optimal patient-specific therapies. (b) Schematic representation of PDLS evaluation. At the endpoint, the spheroid is picked up and digested into a single cell suspension (3D-in). The remaining cells and culture media in the plate-well form the 3D-out compartment. The amount of leukaemia cells is evaluated in the 3D-in and -out by flow cytometry. (c) Visualization of a co-culture after 24 h using an EVOS microscope system. Magnification: 4×. Bar: 1000 μm. Unlabelled stromal spheroid in grey. CellTrace violet (CTV) dye-stained leukaemia cells in violet. The flow cytometry analysis allows us to quantify the amount of leukaemia cells that colonize the spheroid (3D-in) and the non-colonizing cells (3D-out). CTV and CD45 were used to distinguish leukaemia cells from the stroma.

12 |. CONCLUSION

Because the BM niche is composed of different components contributing to HSC function, we have described how the disruption of these components can favour the maintenance of LSCs at expense of HSCs. The specific interdependence between LSCs and BM niches is summarized in Figure 1. As the understanding the mechanisms involved in such relationships is improved, the development of combinatorial therapies that effectively ablate LSCs within the BM niche will also be improved. Future work in reproducing in vitro the crosstalk between LSCs and the human BM niche is needed to create ex vivo models to evaluate new therapies.

Abbreviations:

ALL

acute lymphoblastic leukaemia

AML

acute myeloid leukaemia

CAR

C-X-C motif chemokine ligand 12-abundant reticular

CML

chronic myeloid leukaemia

FLT3-ITD

fms-like tyrosine kinase 3-internal tandem duplication

GAT

gonadal adiposity tissue

HA

hyaluronan

HSCs

haematopoietic stem cells

HSPCs

haematopoietic stem/progenitor cells

ICOS

inducible T cell co-stimulator

ICOSL

inducible T cell co-stimulator ligand

LepR

leptin receptor

LSCs

leukaemia stem cells

MSCs

mesenchymal stem/stromal cells

PDLS

patient-derived leukaemic spheroids

Runx

runt-related transcription factor

VCAM

vascular cell adhesion molecule

VLA

very late antigen