The bone marrow niche from the inside out: how megakaryocytes are shaped by and shape hematopoiesis

Platelets are small anucleate discoid cells that range from 1 to 3 µm in diameter. They are the most critical players in maintaining hemostasis and vascular integrity and have key roles in thrombosis and inflammatory responses. Platelets are formed from megakaryocytes (MKs), their precursor cells, which reside predominantly in the bone marrow (BM). MKs are large hematopoietic cells, making up ≈0.05% of the cell population in human BM. After MKs are terminally differentiated, they undergo a complex maturation process during which they increase in size, become full of platelet-specific granules, expand their cytoplasmic content of cytoskeletal proteins, and develop a highly tortuous invaginated (demarcation) membrane system.¹ 

Megakaryopoiesis is the process by which MKs are derived from hematopoietic stem cells (HSCs). HSCs sit atop a hierarchy of progenitors that undergo consecutive commitment steps, becoming progressively restricted in their lineage potential and finally producing unipotent progenitors.² According to the classical model of hematopoiesis, each mature MK is derived from an HSC that sequentially transitions through the multipotent progenitor (MPP), common myeloid progenitor (CMP), MK-erythroid progenitor (MEP), and MK progenitor (MkP) states.³ During the last decade, this developmental structure has been disputed, with the origin of MK precursors being one of the most actively debated subjects. Recent studies have suggested that MKs are generated from multiple pathways, some of which do not require transit through multipotent or bipotent MEP stages in steady-state and emergency conditions.^4,5 Hematopoietic stem and progenitor cells (HSPCs) reside in specialized local microenvironments within the BM, known as BM niches,⁶ that provide crucial signals to regulate HSPC survival, quiescence, mobilization, and differentiation. Growing evidence suggests that stress-induced molecular changes in the BM niche directly and/or indirectly regulate HSC activation, expansion, and differentiation.⁷ In this review, we explore how the BM niche can encourage and alter MK development.

MKs, the largest of the hematopoietic cells, assemble and release platelets by extending long proplatelet structures into the bloodstream. Because of the inaccessibility of the BM, the cellular and molecular basis of how MKs give birth to platelets has historically been poorly understood. However, the purification and cloning of the glycoprotein hormone thrombopoietin (TPO) in 1994 allowed the first insights into the differentiation of these unique cells.⁸ TPO is the principal regulator of megakaryopoiesis, regulating all stages of MK differentiation, from the HSC through cytoplasmic maturation.^9,10 TPO has also been linked to regulation of HSC survival and proliferation.^11,12 Nonetheless, mice lacking TPO or its receptor c-Mpl have platelets and MKs, although at severely reduced levels (thrombocytopenia), suggesting that TPO signaling is not the sole regulator of HSC differentiation into MKs.^11-13 In addition, although TPO directs the commitment of HSCs to the MK lineage and controls MK maturation, it does not regulate the production of platelets from MKs.² During the final stages of maturation, the MK cytoplasm undergoes massive cytoskeletal remodeling to induce the formation of proplatelets, which are shed into the bloodstream, where they fracture into platelets.^14,15 Of note, our understanding of what triggers mature MKs to begin the process of proplatelet formation is still very limited.

Despite tremendous diversity in cell morphology and function, blood and immune cells are all derived from multipotent HSCs.¹⁶ In the classical model of hematopoiesis, HSCs sit atop of a developmental hierarchy and give rise to increasingly restricted progenitors that eventually produce, and continuously replenish, ≈1 trillion mature blood cells per day.¹⁷ In this model, long-term repopulating HSCs (LT-HSCs) give rise to short-term repopulating HSCs (ST-HSCs). These ST-HSCs have a reconstitution ability of <1 month upon transplantation, after which they differentiate into MPPs with no self-renewal capacity.¹⁸ MPPs (MPP2-4) have been designated by researchers, such as Pietras et al, as a set of unique signature and lineage-biased progenitors independently generated by HSCs; MPP1-4 are unable to generate other MPPs in vivo after transplantation.¹⁹ MPP2 is biased toward the MK/erythroid lineage, MPP3 is largely granulocyte/macrophage biased, and MPP4 is largely lymphoid biased.¹⁹ Recently, an additional population, MPP5, has been described in mice; these cells can generate MPP1-4 but not HSCs.²⁰ As such, MPP5 is located immediately downstream of HSCs but upstream of the more committed MPP2-4.²⁰ MPPs differentiate into common lymphoid progenitors, which display lymphoid potential and differentiate into mature B and T cells, and CMPs that generate granulocyte-macrophage progenitors and MEPs. MEPs then give rise to unipotent MkPs, which mature into MKs (Figure 1A).² 

Over the past decade, the concept of a hierarchical lineage tree and stable discrete HSC populations has been revised into a more dynamic model.²¹ These findings are now supported by advances in single-cell omics analyses, which are further unraveling novel HSC subpopulations.²² In particular, the origin of MKs has been under debate; recent findings revealed similarities between HSCs and MKs, pointing toward the possibility that the MK pathway is bifurcated earlier than assumed in the classical model.^19,23 In 2007, Månsson et al described an MK-like subset of myeloid-biased HSCs using global and single-cell genomics tools.⁴ Under steady-state conditions, they found HSCs with high expression of MK-associated genes (Gata1, Epor, Mpl, and vWF). Further evidence for MK-primed HSCs was reported simultaneously by 2 groups. Gekas and Graf identified a CD41⁺ subset of myeloid-biased HSCs that becomes more prevalent with age,²⁴ and Jacobsen and colleagues found that LT-HSCs expressing vWF are a TPO-dependent HSC population capable of short-term and stable long-term platelet-biased reconstitution after transplantation.²⁵ Shortly after, Park and colleagues revealed that c-Kit^hi cells are also an MK-biased population.²⁶ Intriguingly, Sanjuan-Pla et al showed coexpression of the MK bias markers vWF and CD41 on 67.9% of LSK/CD34⁻/CD150⁺/CD48⁻ cells.²⁵ Moving forward, another marker, GABRR1, was found to be expressed in subsets of HSCs and MkPs, but not in other progenitor populations.²⁷ Collectively, these studies reveal a variety of “MK-biased” HSCs identifiable through distinct marker expression.

Efforts to examine the functionality of MK-biased populations found HSCs that differentiate directly into MK, MK-erythroid, or common myeloid lineages.²¹ Based on single-cell transplantation experiments, a myeloid bypass model of differentiation was proposed, directly branching from the most primitive HSCs. Later, Haas et al reported that the HSC compartment contains stem-like MK-committed progenitors (SL-MkPs) that remain quiescent during steady-state.⁵ Acute inflammatory signaling triggers cell cycle activation of these SL-MkPs, initiating the translation of MK transcripts. The investigators showed that 28 proteins typically expressed in mature MKs and platelets (eg, CD41, vWF, and CXCL4 (platelet factor 4) were upregulated in HSCs upon inflammation, resulting in a rapid maturation of SL-MkPs and other MkPs⁵ (Figure 1A). Similarly, HSCs have the capacity to differentiate directly into MKs in myelofibrosis,²⁸ and the numbers of LT-HSCs, ST-HSCs, and MkPs also increase during aging.^29,30 Grover et al revealed that HSC aging is accompanied by upregulation of platelet-lineage gene expression.³⁰ Further, their study showed that the age-dependent increase in platelet bias correlates with a selective expansion of platelet-primed vWF⁺ HSCs, implicating MK-biased HSCs in the aging platelet phenotype. Taken together, these data suggest that, in contrast to the concept of stable and discrete HSPC populations, a more dynamic relationship exists with plasticity in the proliferation and differentiation capacity of HSCs. Furthermore, there is compelling evidence that HSC subsets can be partially or completely biased toward MK differentiation and platelet production during inflammation. In the context of these findings, it is evident that interactions within the BM niche impact regulation of hematopoiesis and megakaryopoiesis. In the remainder of this review, we uncover some ways in which the BM environment may shape and regulate megakaryopoiesis and how resulting MKs impact the niche.

Organs and tissues are organized to facilitate their function, and the BM is no different. The BM is a highly complex tissue comprising an intricate network of endothelial cells (including sinusoids, arterioles, and transition zone vessels), multipotent mesenchymal stromal cells (MSCs) and their progeny (osteoblasts, chondrocytes, and adipocytes), and HSPCs and their progeny, such as MKs and macrophages.^31,32 The term “niche,” first coined by Schofield in 1978, refers to the environment in which a cell can be found that is fundamental to its survival and function.³³ It has been well documented that the BM niche can be extensively remodeled in a variety of conditions, such as inflammation,⁵ infection,³⁴ radiation therapy,³⁵ and aging,^36,37 which has a propagating effect on HSPC differentiation and development. Understanding the mechanisms by which the HSC niche is regulated can be crucial to improving treatment outcomes and eradicating disease.³⁸ For example, Psaila et al used single-cell omics to demonstrate MK-biased hematopoiesis in myelofibrosis propagated by HSPCs with aberrant expression of G6B on their surface.²⁸ This G6B expression can now be exploited as a selective immunotherapeutic target in the clinic.

BM niche changes result from insults that affect factors such as BM cellular composition, local microenvironment, architecture, or a combination of these factors. The effectors of these changes are broad and range from physical insults, such as radiation, to signaling induced by secreted factors in the blood. In this section, we expand on how 3 components, direct cell-to-cell contact, secreted molecules (autocrine and paracrine signaling), and the release of cellular components like extracellular vesicles (EVs), impact MKs and the BM niche.

HSCs rely on interactions in their niche with stroma, endothelium, and hematopoietic cells that facilitate their survival and regulate their function and differentiation.^39-42 Building on this and other early observations, the HSC niche is now viewed as a complex assembly of multiple components existing in close proximity to each other that provide molecular and physical cues that regulate the quiescence, proliferation, and differentiation of HSCs.⁴³ Within the field examining the BM microenvironment, considerable focus has been placed on the development of methods to identify the in situ localization of HSCs and other cellular components that contribute to their niche. For example, a study of myelopoietic organization in murine sternal marrow identified a predominantly sinusoidal localization for myeloid lineage cells and progenitors.⁴⁴ This study demonstrated coordination of myelopoiesis in sinusoidal niches by colony-stimulating factor 1 expression in selected vessels.⁴⁴ Better knowledge of the constituents of the MK niche and the cells with which they interact may improve our understanding of megakaryopoiesis during normal and disease states.

Despite representing a minor proportion of hematopoietic cells in the BM, the large size of MKs and their ubiquitous distribution suggest the potential to interact with many cells. Using a whole-mount 3-dimensional immunofluorescence imaging technique, Bruns et al revealed that nearly 30% of HSCs are found within 5 μm of MKs, a higher ratio than would be expected from random simulations.⁴⁵ This was corroborated by a recent study showing that a proportion of α-catulin⁺ HSCs were found within 5 μm of MKs, suggesting a preferential localization of some HSCs near MKs.⁴⁶ Another study indicated that MKs play an important role in the regulation of HSC quiescence by production of transforming growth factor-β1 (TGF-β1) and recovery after stress by transient fibroblast growth factor (FGF) signaling.⁴⁷ Additionally, when MKs are depleted from the BM niche, there is a significant increase in HSC numbers, pointing toward MKs as a potential regulator of HSC quiescence.^45,47-49 Of note, MK depletion is also enough to reprogram vWF⁺ HSCs from myeloid-biased to balanced-lineage contributions.⁴⁹ Mechanistically, because MKs inhibit HSC proliferation through secreted proteins, including TGF-β1⁴⁷ and CXCL4,⁴⁵ direct contact between MKs and HSCs may not be required⁴⁶ (Table 1). CXCL4 is also known to inhibit the proliferation of megakaryocytes and promotes maturation via autocrine signaling.^50,51 Collectively, these data reveal an important functional role for MKs in the BM niche beyond the production of platelets.

Another unique cellular relationship is the engulfment and subsequent release of neutrophils by MKs in a process termed emperipolesis. Observations of neutrophils inside MKs were first reported from electron microscopy images taken in the 1980s.⁵² A more recent investigation highlighted the functional implications of this relationship; an exchange of membrane material between the MK and neutrophil can occur during emperipolesis, with the potential for exchange of other materials during this process.⁵³ This interaction results in accelerated maturity to platelet production by MKs.⁵³ However, the percentage of MKs that undergo this process and the frequency with which it occurs have yet to be reliably determined under normal and stress hematopoiesis, leaving unanswered questions as to the role of this phenomenon for normal and/or pathological MK development.

Given that proplatelet extension through the endothelium into the vascular lumen is required for platelet production, there are undoubtedly many key interactions between MKs and endothelial cells. Penetration into the vessels occurs through podosomes, actin-driven structures that facilitate a transendothelial extension.⁵⁴ Little is known about how MKs determine the polarity of their proplatelet extensions toward endothelial cells and what interactions are necessary to facilitate podosome formation and the elongation of these extensions through the cytoplasm of the endothelial cells. Furthermore, endothelial cells are extremely sensitive and changeable during inflammation, upregulating production of cytokines and adhesion molecules.⁵⁵ For example, Fernandez et al showed that BM endothelial cells promote expansion of HPSCs by a Notch-dependent mechanism upon inflammation.⁵⁵ However, these interactions have been difficult to study, because they require complex coculture systems that recapitulate BM architecture. As such, further studies are needed to understand the complex relationships between endothelial cells and MKs.

BM adipocytes have been described as negative regulators of the hematopoietic environment.⁵⁶ Recent studies have also introduced a potentially key role for lipids from adipocytes in cell fate decisions (Figure 1B). A study carried out by Severin and colleagues⁵⁷ showed that MKs take up fatty acids released by adipocytes via a CD36-dependent mechanism to facilitate their maturation, suggesting that lipids have an essential role in MK development. Further supporting this, Kelly et al demonstrated that fatty acid incorporation through acetyl-CoA carboxylase in the de novo lipogenesis pathway is a key regulator of late-stage MK maturation and platelet formation.⁵⁸ This interaction raises questions about the relationship between megakaryopoiesis and adipocytes in the context of metabolic diseases. For example, the established increase in BM adiposity with age⁵⁹ may result in altered MK or platelet function, which could help explain the platelet hyperreactivity that contributes to the increased risk of heart attack and stroke.

BM-resident cells secrete growth factors, cytokines, and lipids that orchestrate and maintain the hematopoietic response by mediating short- and long-range communication (Table 1).⁶⁰ Therefore, examining these secreted signals is essential in understanding the remodeling of BM niches. Cytokines play a pivotal role in HSC regulation and maintenance and are also required for megakaryopoiesis.⁹ However, little is known about the precise sources of these molecules; they may originate from BM cells, permeate from the blood stream, or derive from a combination of many origins (Figure 1C). The most well-studied is TPO, which is expressed mainly in the liver and accesses the BM via the bloodstream. There, it regulates MK differentiation from HSCs and the subsequent progenitor cells.¹⁰ In addition to MK differentiation, TPO has an important role in HSC maintenance. A study by Decker et al showed that following targeted deletion of TPO from hepatocytes, BM HSCs are depleted, pointing to hepatocyte-derived TPO as the major functional source of TPO for BM HSC maintenance under steady-state conditions.⁶¹ 

Although many investigators have explored how TPO acts on HSC differentiation and MK maturation,² few studies have addressed the effect of other cytokines and their impact on megakaryopoiesis. Levels of CCL5, or RANTES, increase in the BM microenvironment with aging, which is associated with a positive bias in myeloid cell production.⁶² Accordingly, exposure to CCL5 or constitutive expression of CCL5 results in lineage skewing to favor production of myeloid cells and a corresponding decrease in lymphocyte production.⁶² Recent studies demonstrate that CCL5 could be used pharmacologically following therapeutic or accidental radiation injury as a result of its role in hematopoietic regeneration.⁶³ These findings agree with previous studies from our laboratory that found CCL5 signaling through its receptor CCR5 enhanced MK ploidy and proplatelet formation through prosurvival signaling downstream of BCL2-associated agonist of cell death phosphorylation.⁶⁴ Together, these studies suggest that CCL5 may bias HSCs toward the megakaryocytic lineage and subsequently enhance platelet production from mature MKs.

Members of the interleukin (IL) family can also directly influence HSCs, driving their proliferation and skewing them toward the myeloid lineage.^65,66 IL-1β and IL-6 increase MK maturation and, consequently, platelet counts.³⁷ However, their mechanisms of action were ultimately discovered to be predominantly through an indirect effect of increasing TPO production.² Conversely, IL-1α, which is involved in inflammatory processes and hematopoiesis, has been reported to drive TPO-independent differentiation of MKs.⁶⁷ Nishimura et al further reported that IL-1α induces enhanced platelet release through rupture of the mature MK membrane.⁶⁸ 

Stromal cell derived factor-1/CXCL12 was the first chemokine to be implicated in MK maturational chemotaxis to the vascular niche,⁶⁹ a concept that is now being challenged. CXCL12 is produced by several cell types in the BM, such as osteoblasts, endothelial cells, and perivascular MSCs. Niswander et al demonstrated that CXCL12 regulates the spatial distribution of MKs in the marrow vasculature,⁷⁰ pointing toward the importance of CXCL12 in the organization of the BM niche. Moreover, a recent study found that β4GalT1, regulated by CXCL12, controls β1 integrin function in MKs and HSC homeostasis.⁷¹ Similar to CXCL12, administration of FGF4 increases platelet counts in vivo and partially rescues thrombopoiesis in Tpo^−/− and Mpl^−/− mice by directly augmenting cellular adhesion of MKs to endothelial cells.⁷² 

The capacity for cytokine signaling within the BM can be regulated to some degree by extracellular matrix (ECM) components distributed throughout the BM niche. ECM is composed of various proteins and glycoproteins that provide structure to the niche, as well as mechanical and biochemical signals to cells in the BM environment, including MKs.⁷³ During megakaryopoiesis, ECM components, such as type-1 collagen, fibronectin, and fibrinogen, directly affect MK development and platelet production in the BM.^74,75 In accordance, dysregulation of ECM production and remodeling is associated with BM pathologies. For example, during myelofibrosis, secretion of TGF-β1 by MKs results in overproduction of ECM by stroma and subsequent restriction of BM function.⁷⁶ 

Lipids are another class of molecules that regulate many vital processes in hematopoiesis and homing of HSPCs in the BM niche. However, less is known about the effects of bioactive lipids and lipid mediators on regulating hematopoiesis and, by extension, megakaryopoiesis. During MK maturation, MKs develop a complex demarcation membrane system that becomes enriched with the plasma membrane lipids phosphoinositide and phosphatidylinositol 4,5-bisphosphate, which represent markers for internal membranes of mature MKs.⁷⁷ Similarly, phosphatidylinositol 3-monophosphate is a key component in vesicular trafficking and autophagy. Of note, abnormal autophagy is associated with impaired MK differentiation and platelet production ( see Sun et al⁷⁸ for additional information). Recently, phosphatidylinositol 3-monophosphate, produced by Vps34 (phosphatidylinositol 3-kinase), has emerged as an important regulator of MK/platelet structure and function⁷⁹; inhibiting Vps34 at early stages of MK development leads to aberrant demarcation membrane system development.⁸⁰ Another important lipid, platelet activation factor (PAF), is present in the BM and regulates the function of HSPCs. PAF diminishes the growth of human myeloid and erythroid progenitors, and PAF dysregulation in the BM may influence trafficking, activation, and gene expression patterns in neighboring cells.^81,82 In addition, sphingosine-1-phosphate (S1P) regulates important biological functions, including cell migration, in the BM compartment.⁸³ S1P also induces CXCL12 secretion from MSC populations in the BM niche.⁸⁴ In the context of megakaryopoiesis, S1P and the MK S1P receptor (S1pr1) are indispensable for normal BM thrombopoiesis, and Niazi et al proposed that S1P1 serves as a critical directional cue guiding the elongation of proplatelet extensions from the interstitium into BM sinusoids.⁸⁵ Moreover, the investigators also found that S1P restricts megakaryopoiesis through S1pr1 and can further suppress thrombopoiesis through S1pr2 when aberrantly secreted in the hematopoietic niche,⁸⁵ suggesting a more complex role for this signaling pathway. In sum, there is emerging literature indicating an important role for secreted molecules in HSC differentiation and MK maturation. Further knowledge of the microenvironmental regulation of the MK lineage could lead to improved therapeutic strategies that can specifically enhance or repress MK differentiation from HSCs.

Interest in EVs has increased dramatically as a result of their implication in cell-to-cell communication, disease development and progression, and candidacy as potential therapeutics. EVs are membrane-enclosed structures of varying size (30-10 000 nm) that are released from cells and mediate local and distant intercellular communication.⁸⁶ Indeed, EVs of various origins have been identified in every biofluid tested to date, with substantial variation in their size, structure, content, and function.⁸⁷ EVs bind to and are internalized by cells to alter cell function.^88-93 Membrane fusion and/or internalization of EVs allow EV membrane and cargo to be transferred into the recipient cell cytoplasm; lipids, RNA, and protein are delivered by these mechanisms to target cells.^94,95 As such, EVs can be messengers of diverse cargo, reflecting the local environment from which they originate. Growing evidence supports the importance of EVs released from BM cells in communication within the BM microenvironment.⁹⁶ 

EVs derived from MKs (MKEVs) and EVs derived from platelets (PEVs) are the most abundant in circulation, accounting for almost 80% of the total population.^97-100 PEVs are released upon platelet stimulation and are generally characterized as CD41⁺/CD62P⁺, whereas MKEVs are constitutively released and are generally CD41⁺/CD62P⁻.¹⁰¹ Within the BM, MKEVs interact with HSPCs through receptors such as ICAM-1, CD43, CD18, and CD11b.⁹¹ Functionally, the internalization of MKEVs redirects the differentiation of HSPCs toward MKs, with limited effects on the phenotype of endothelial or stromal cells.⁹¹ The mechanism of this reprogramming was found to be through RNA in MKEVs, later suggested to be miR-1915-3p.⁹¹ 

In addition, our group recently demonstrated that PEVs leave the circulation and infiltrate the BM, where they bind to resident BM cells, including HSCs and MKs, and can affect their development (Figure 1D).¹⁰¹ This finding introduces the idea that EVs may be communicating from cells within the BM, as well as from cells at distant sites. Circulating platelets are known to endocytose many components of plasma; thus, during disease their content changes to reflect the disease state. These findings suggest a model whereby PEVs can penetrate the BM and relay pathologies from the plasma directly to the BM microenvironment.

The study of HSCs in the BM niche has traditionally been dominated by different concepts: (1) a static long-term HSC responsible for maintaining hematopoiesis, (2) a model of progenitor hierarchy in which HSCs differentiate into increasingly committed progenitors and, subsequently, mature lineages along nonoverlapping paths, and (3) an HSC niche, where HSCs reside in discrete anatomical microenvironments. However, during the past decade it has been shown that the BM niche may be more dynamic and complicated, especially once activated by conditions like inflammation.

Moving forward, studies continuing to explore the roles that MKs play in reshaping the BM niche will be key in understanding hematopoiesis during health and disease. For example, although committed MkPs have been reported by many investigators to expand during BM stress, the specific molecules and signaling pathways responsible for this are still unknown. In addition, most of the in vitro experiments discussed in this review were done in 2-dimensional culture systems, which do not take into account BM architecture. The use of biochips and 3-dimensional culture systems will be essential in helping to gain a better understanding of how the complex relationships of cells and matrix within the BM niche create and shape the local environment.

The authors thank Kellie R. Machlus for her support in this review and Kristin Johnson for producing the figures.

M.N.B. is supported by grants from the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases K01DK111515 and R03DK124746 and National Heart, Lung, and Blood Institute R01HL151494.

M.N.B designed, wrote, and edited the manuscript; A.P.S. wrote and revised the manuscript; and T.F.N. wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Maria N. Barrachina, Karp Family Research Building, 1 Blackfan Circle, Room 11.004.H, Boston, MA 02115; e-mail: maria.barrachina@childrens.harvard.edu.